CLOUD computing is quickly becoming an effective and

Size: px
Start display at page:

Download "CLOUD computing is quickly becoming an effective and"

Transcription

1 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 24, NO. 6, JUNE Otial Multiserver Configuration for Profit Maxiization in Cloud Couting Junwei Cao, Senior Meber, IEEE, Kai Hwang, Fellow, IEEE, Keqin Li, Senior Meber, IEEE, and Albert Y. Zoaya, Fellow, IEEE Abstract As cloud couting becoes ore and ore oular, understanding the econoics of cloud couting becoes critically iortant. To axiize the rofit, a service rovider should understand both service charges and business costs, and how they are deterined by the characteristics of the alications and the configuration of a ultiserver syste. The roble of otial ultiserver configuration for rofit axiization in a cloud couting environent is studied. Our ricing odel takes such factors into considerations as the aount of a service, the workload of an alication environent, the configuration of a ultiserver syste, the service-level agreeent, the satisfaction of a consuer, the quality of a service, the enalty of a low-quality service, the cost of renting, the cost of energy consution, and a service rovider s argin and rofit. Our aroach is to treat a ultiserver syste as an M/M/ queuing odel, such that our otiization roble can be forulated and solved analytically. Two server seed and ower consution odels are considered, naely, the idle-seed odel and the constant-seed odel. The robability density function of the waiting tie of a newly arrived service request is derived. The exected service charge to a service request is calculated. The exected net business gain in one unit of tie is obtained. Nuerical calculations of the otial server size and the otial server seed are deonstrated. Index Ters Cloud couting, ultiserver syste, ricing odel, rofit, queuing odel, resonse tie, server configuration, service charge, service-level agreeent, waiting tie Ç INTRODUCTION CLOUD couting is quickly becoing an effective and efficient way of couting resources and couting services consolidation [0]. By centralized anageent of resources and services, cloud couting delivers hosted services over the Internet, such that accesses to shared hardware, software, databases, inforation, and all resources are rovided to consuers on-deand. Cloud couting is able to rovide the ost cost-effective and energy-efficient way of couting resources anageent and couting services rovision. Cloud couting turns inforation technology into ordinary coodities and utilities by using the ay-er-use ricing odel [3], [5], [8]. However, cloud couting will never be free [8], and understanding the econoics of cloud couting becoes critically iortant. One attractive cloud couting environent is a threetier structure [5], which consists of infrastructure vendors, service roviders, and consuers. The three arties are also called cluster nodes, cluster anagers, and consuers. J. Cao is with the Research Institute of Inforation Technology, Tsinghua National Laboratory for Inforation Science and Technology, Tsinghua University, Beijing 00084, China. E-ail: jcao@tsinghua.edu.cn.. K. Hwang is with the Deartent of Electrical Engineering, University of Southern California, Los Angeles, CA E-ail: kaihwang@usc.edu.. K. Li is with the Deartent of Couter Science, State University of New York, New Paltz, New York 256. E-ail: lik@newaltz.edu.. A.Y. Zoaya is with the School of Inforation Technologies, University of Sydney, Sydney, NSW 2006, Australia. E-ail: albert.zoaya@sydney.edu.au. Manuscrit received 23 Feb. 202; revised 8 June 202; acceted 28 June 202; ublished online 25 June 202. Recoended for accetance by V.B. Misic, R. Buyya, D. Milojicic, and Y. Cui. For inforation on obtaining rerints of this article, lease send e-ail to: tds@couter.org, and reference IEEECS Log Nuber TPDSSI Digital Object Identifier no. 0.09/TPDS in cluster couting systes [2], and resource roviders, service roviders, and clients in grid couting systes [9]. An infrastructure vendor aintains basic hardware and software facilities. A service rovider rents resources fro the infrastructure vendors, builds aroriate ultiserver systes, and rovides various services to users. A consuer subits a service request to a service rovider, receives the desired result fro the service rovider with certain service-level agreeent, and ays for the service based on the aount of the service and the quality of the service. A service rovider can build different ultiserver systes for different alication doains, such that service requests of different nature are sent to different ultiserver systes. Each ultiserver syste contains ultile servers, and such a ultiserver syste can be devoted to serve one tye of service requests and alications. An alication doain is characterized by two basic features, i.e., the workload of an alication environent and the exected aount of a service. The configuration of a ultiserver syste is characterized by two basic features, i.e., the size of the ultiserver syste (the nuber of servers) and the seed of the ultiserver syste (execution seed of the servers). Like all business, the ricing odel of a service rovider in cloud couting is based on two coonents, naely, the incoe and the cost. For a service rovider, the incoe (i.e., the revenue) is the service charge to users, and the cost is the renting cost lus the utility cost aid to infrastructure vendors. A ricing odel in cloud couting includes any considerations, such as the aount of a service (the requireent of a service), the workload of an alication environent, the configuration (the size and the seed) of a ultiserver syste, the service-level agreeent, the satisfaction of a consuer (the exected service tie), the quality of a service (the task waiting tie and the task resonse tie), /3/$3.00 ß 203 IEEE Published by the IEEE Couter Society

2 088 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 24, NO. 6, JUNE 203 the enalty of a low-quality service, the cost of renting, the cost of energy consution, and a service rovider s argin and rofit. The rofit (i.e., the net business gain) is the incoe inus the cost. To axiize the rofit, a service rovider should understand both service charges and business costs, and in articular, how they are deterined by the characteristics of the alications and the configuration of a ultiserver syste. The service charge to a service request is deterined by two factors, i.e., the exected length of the service and the actual length of the service. The exected length of a service (i.e., the exected service tie) is the execution tie of an alication on a standard server with a baseline or reference seed. Once the baseline seed is set, the exected length of a service is deterined by a service request itself, i.e., the service requireent (aount of service) easured by the nuber of instructions to be executed. The longer (shorter, resectively) the exected length of a service is, the ore (less, resectively) the service charge is. The actual length of a service (i.e., the actual service tie) is the actual execution tie of an alication. The actual length of a service deends on the size of a ultiserver syste, the seed of the servers (which ay be faster or slower than the baseline seed), and the workload of the ultiserver syste. Notice that the actual service tie is a rando variable, which is deterined by the task waiting tie once a ultiserver syste is established. There are any different service erforance etrics in service-level agreeents [2]. Our erforance etric in this aer is the task resonse tie (or the turn around tie), i.e., the tie taken to colete a task, which includes task waiting tie and task execution tie. The service-level agreeent is the roised tie to colete a service, which is a constant ties the exected length of a service. If the actual length of a service is (or, a service request is coleted) within the service-level agreeent, the service will be fully charged. However, if the actual length of a service exceeds the service-level agreeent, the service charge will be reduced. The longer (shorter, resectively) the actual length of a service is, the ore (less, resectively) the reduction of the service charge is. In other words, there is enalty for a service rovider to break a service-level agreeent. If the actual service tie exceeds certain liit (which is service request deendent), a service will be entirely free with no charge. Notice that the service charge of a service request is a rando variable, and we are interested in its exectation. The cost of a service rovider includes two coonents, i.e., the renting cost and the utility cost. The renting cost is roortional to the size of a ultiserver syste, i.e., the nuber of servers. The utility cost is essentially the cost of energy consution and is deterined by both the size and the seed of a ultiserver syste. The faster (slower, resectively) the seed is, the ore (less, resectively) the utility cost is. To calculate the cost of energy consution, we need to establish certain server seed and ower consution odels. To increase the revenue of business, a service rovider can construct and configure a ultiserver syste with any servers of high seed. Since the actual service tie (i.e., the task resonse tie) contains task waiting tie and task execution tie, ore servers reduce the waiting tie and faster servers reduce both waiting tie and execution tie. Hence, a owerful ultiserver syste reduces the enalty of breaking a service-level agreeent and increases the revenue. However, ore servers (i.e., a larger ultiserver syste) increase the cost of facility renting fro the infrastructure vendors and the cost of base ower consution. Furtherore, faster servers increase the cost of energy consution. Such increased cost ay counterweight the gain fro enalty reduction. Therefore, for an alication environent with secific workload which includes the task arrival rate and the average task execution requireent, a service rovider needs to decide an otial ultiserver configuration (i.e., the size and the seed of a ultiserver syste), such that the exected rofit is axiized. In this aer, we study the roble of otial ultiserver configuration for rofit axiization in a cloud couting environent. Our aroach is to treat a ultiserver syste as an M/M/ queuing odel, such that our otiization roble can be forulated and solved analytically. We consider two server seed and ower consution odels, naely, the idle-seed odel and the constant-seed odel. Our ain contributions are as follows. We derive the robability density function (df) of the waiting tie of a newly arrived service request. This result is significant in its own right and is the base of our discussion. We calculate the exected service charge to a service request. Based on these results, we get the exected net business gain in one unit of tie, and obtain the otial server size and the otial server seed nuerically. To the best of our knowledge, there has been no siilar investigation in the literature, although the ethod of otial ulticore server rocessor configuration has been eloyed for other uroses, such as anaging the ower and erforance tradeoff [7]. One related research is user-centric and arket-based and utility-driven resource anageent and task scheduling, which have been considered for cluster couting systes [7], [20], [2] and grid couting systes [4], [2], [9]. To coete and bid for shared couting resources through the use of econoic echaniss such as auctions, a user can secify the value (utility, yield) of a task, i.e., the reward (rice, rofit) of coleting the task. A utility function, which easures the value and iortance of a task as well as a user s tolerance to delay and sensitivity to quality of service, suorts arket-based bidding, negotiation, and adission control. By taking an econoic aroach to roviding service-oriented and utility couting, a service rovider allocates resources and schedules tasks in such a way that the total rofit earned is axiized. Instead of traditional syste-centric erforance otiization such as iniizing the average task resonse tie, the ain concern in such coutational econoy is user-centric erforance otiization, i.e., axiizing the total utility delivered to the users (i.e., the total user-erceived value). 2 AMULTISERVER MODEL Throughout the aer, we use P½eŠ to denote the robability of an event e. For a rando variable x, we use f x ðtþ to reresent the robability density function of x, and F x ðtþ to

3 CAO ET AL.: OPTIMAL MULTISERVER CONFIGURATION FOR PROFIT MAXIMIZATION IN CLOUD COMPUTING 089 reresent the cuulative distribution function (cdf) of x, and x to reresent the exectation of x. A cloud couting service rovider serves users service requests by using a ultiserver syste, which is constructed and aintained by an infrastructure vendor and rented by the service rovider. The architecture detail of the ultiserver syste can be quite flexible. Exales are blade servers and blade centers where each server is a server blade [6], clusters of traditional servers where each server is an ordinary rocessor [7], [20], [2], and ulticore server rocessors where each server is a single core [7]. We will sily call these blades/rocessors/cores as servers. Users (i.e., custoers of a service rovider) subit service requests (i.e., alications and tasks) to a service rovider, and the service rovider serves the requests (i.e., run the alications and erfor the tasks) on a ultiserver syste. Assue that a ultiserver syste S has identical servers. In this aer, a ultiserver syste is treated as an M/M/ queuing syste which is elaborated as follows. There is a Poisson strea of service requests with arrival rate, i.e., the interarrival ties are indeendent and identically distributed (i.i.d.) exonential rando variables with ean =. A ultiserver syste S aintains a queue with infinite caacity for waiting tasks when all the servers are busy. The first-coe-first-served (FCFS) queuing disciline is adoted. The task execution requireents (easured by the nuber of instructions to be executed) are i.i.d. exonential rando variables r with ean r. The servers (i.e., blades/rocessors/cores) of S have identical execution seed s (easured by the nuber of instructions that can be executed in one unit of tie). Hence, the task execution ties on the servers of S are i.i.d. exonential rando variables x ¼ r=s with ean x ¼ r=s. Notice that although an M/G/ queuing syste has been considered (see, e.g., [3]), the M/M/ queuing odel is the only odel that accoodates an analytical and closedfor exression of the robability density function of the waiting tie of a newly arrived service request. Let ¼ =x ¼ s=r be the average service rate, i.e., the average nuber of service requests that can be finished by a server of S in one unit of tie. The server utilization is ¼ = ¼ x= ¼ = r=s, which is the average ercentage of tie that a server of S is busy. Let k denote the robability that there are k service requests (waiting or being rocessed) in the M/M/ queuing syste for S. Then, we have ([4,. 02]) 8 >< k ¼ ðþ k 0 ; k ; k! >: k 0 ; k ;! where 0 ¼ X k¼0 ðþ k k! þ ðþ!! : The robability of queuing (i.e., the robability that a newly subitted service request ust wait because all servers are busy) is P q ¼ X k¼ k ¼ ¼ ðþ 0! : The average nuber of service requests (in waiting or in execution) in S is N ¼ X k¼0 k k ¼ þ P q: Alying Little s result, we get the average task resonse tie as T ¼ N! ¼ x þ P q ¼ x þ ð Þ ð Þ 2 : The average waiting tie of a service request is W ¼ T x ¼ ð Þ 2 x: The waiting tie is the source of custoer dissatisfaction. A service rovider should kee the waiting tie to a low level by roviding enough servers and/or increasing server seed, and be willing to ay back to a custoer in case the waiting tie exceeds certain liit. 3 POWER CONSUMPTION MODELS Power dissiation and circuit delay in digital CMOS circuits can be accurately odeled by sile equations, even for colex icrorocessor circuits. CMOS circuits have dynaic, static, and short-circuit ower dissiation; however, the doinant coonent in a well-designed circuit is dynaic ower consution P (i.e., the switching coonent of ower), which is aroxiately P ¼ acv 2 f, where a is an activity factor, C is the loading caacitance, V is the suly voltage, and f is the clock frequency [6]. In the ideal case, the suly voltage and the clock frequency are related in such a way that V / f for soe constant >0 [22]. The rocessor execution seed s is usually linearly roortional to the clock frequency, naely, s / f. For ease of discussion, we will assue that V ¼ bf and s ¼ cf, where b and c are soe constants. Hence, we know that ower consution is P ¼ acv 2 f ¼ ab 2 Cf 2þ ¼ðab 2 C=c 2þ Þs 2þ ¼ s, where ¼ ab 2 C=c 2þ and ¼ 2 þ. For instance, by setting b ¼ :6, ac ¼ 7:0, c ¼ :0, ¼ 0:5, ¼ 2 þ ¼ 2:0, and ¼ ab 2 C=c ¼ 9:492, the value of P calculated by the equation P ¼ acv 2 f ¼ s is reasonably close to that in [] for the Intel Pentiu M rocessor. We will consider two tyes of server seed and ower consution odels. In the idle-seed odel, a server runs at zero seed when there is no task to erfor. Since the ower for seed s is s, the average aount of energy consued by a server in one unit of tie is s ¼ rs, where we notice that the seed of a server is zero when it is idle. The average aount of energy consued by an -server syste S in one unit of tie, i.e., the ower suly to the ultiserver syste S, isp ¼ s ¼ rs, where ¼ x is the average nuber of busy servers in S. Since a server still consues soe aount of ower P even when it is idle (assue that an idle server consues certain base ower P, which includes static ower dissiation, short-circuit ower dissiation, and other leakage and wasted ower []), we will include P in P, i.e., P ¼ ðs þ P Þ¼rs þ P. Notice that when P ¼ 0, the above P is indeendent of.

4 090 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 24, NO. 6, JUNE 203 In the constant-seed odel, all servers run at the seed s even if there is no task to erfor. Again, we use P to reresent the ower allocated to ultiserver syste S. Since the ower for seed s is s, the ower allocated to ultiserver syste S is P ¼ ðs þ P Þ. 4 WAITING TIME DISTRIBUTION Let W denote the waiting tie of a new service request that arrives to a ultiserver syste. In this section, we find the df f W ðtþ of W. To this end, we consider W in different situations, deending on the nuber of tasks in the queuing syste when a new service request arrives. Let W k denote the waiting tie of a new task that arrives to an M/M/ queuing syste under the condition that there are k tasks in the queuing syste when the task arrives. We define a unit iulse function u z ðtþ as follows: 8 >< u z ðtþ ¼ z; 0 t z ; >: 0; t > z : The function u z ðtþ has the following roerty: Z 0 u z ðtþdt ¼ ; naely, u z ðtþ can be treated as a df of a rando variable with exectation Z 0 tu z ðtþdt ¼ z Z =z 0 tdt ¼ 2z : Let z!and define uðtþ ¼li z! u z ðtþ.it is clear that any rando variable whose df is uðtþ has exectation 0. The following theore gives the df of the waiting tie of a newly arrived service request: Theore. The df of the waiting tie W of a newly arrived service request is f W ðtþ ¼ð P q ÞuðtÞþ e ð Þt ; where P q ¼ =ð Þ and ¼ 0 ðþ =!. Sketch of the Proof. Let W k be the waiting tie of a new service request if there are k tasks in the queuing syste when the service request arrives. We find the df of W k for all k 0. Then, we have f W ðtþ ¼ X k¼0 k f Wk ðtþ: Actually, W k can be found for two cases, i.e., when k< and when k. A colete roof of the theore is given in Section 9. tu Notice that a ultiserver syste with ultile identical servers has been configured to serve requests fro certain alication doain. Therefore, we will only focus on task waiting tie in a waiting queue and do not consider other sources of delay, such as resource allocation and rovision, virtual achine instantiation and deloyent, and other overhead in a colex cloud couting environent. 5 SERVICE CHARGE If all the servers have a fixed seed s, the execution tie of a service request with execution requireent r is known as x ¼ r=s. The resonse tie to the service request is T ¼ W þ x ¼ W þ r=s. The resonse tie T is related to the service charge to a custoer of a service rovider in cloud couting. To study the exected service charge to a custoer, we need a colete secification of a service charge based on the aount of a service, the service-level agreeent, the satisfaction of a consuer, the quality of a service, the enalty of a low-quality service, and a service rovider s argin and rofit. Let s 0 be the baseline seed of a server. We define the service charge function for a service request with execution requireent r and resonse tie T to be 8 ar; if 0 T c r; s 0 ar d T c >< r ; s 0 Cðr; TÞ ¼ if c r<t a s 0 d þ c r; s 0 0; if T> a d þ c >: r: s 0 The above function is defined with the following rationals:. If the resonse tie T to rocess a service request is no longer than ðc=s 0 Þr ¼ cðr=s 0 Þ(i.e., a constant c ties the task execution tie with seed s 0 ), where the constant c is a araeter indicating the servicelevel agreeent, and the constant s 0 is a araeter indicating the exectation and satisfaction of a consuer, then a service rovider considers that the service request is rocessed successfully with high quality of service and charges a custoer ar, which is linearly roortional to the task execution requireent r(i.e., the aount of service), where a is the service charge er unit aount of service (i.e., a service rovider s argin and rofit).. If the resonse tie T to rocess a service request is longer than ðc=s 0 Þr but no longer than ða=d þ c=s 0 Þr, then a service rovider considers that the service request is rocessed with low quality of service and the charge to a custoer should decrease linearly as T increases. The araeter d indicates the degree of enalty of breaking the service-level agreeent.. If the resonse tie T to rocess a service request is longer than ða=d þ c=s 0 Þr, then a service rovider considers that the service request has been waiting too long, so there is no charge and the service is free. Notice that the task resonse tie T is coared with the task execution tie on a server with seed s 0 (i.e., the baseline or reference seed). The actual seed s of a server can be decided by a service rovider, which can be either lower or higher than s 0, deending on the workload (i.e., and r) and syste araeters (e.g.,,, and P ) and the service charge function (i.e., a, c, and d), such that the net business gain to be defined below is axiized.

5 CAO ET AL.: OPTIMAL MULTISERVER CONFIGURATION FOR PROFIT MAXIMIZATION IN CLOUD COMPUTING 09 Fig.. Service charge CðrÞ versus r and. Fig. 2. Noralized service charge CðrÞ=ar versus r and. To build our discussion uon our earlier result on task waiting tie, we notice that the service charge function can be rewritten equivalently in ters of r and W as 8 ar; if 0 W c r; s 0 s a þ cd d >< r dw; s Cðr; WÞ ¼ 0 s c if r<w a s 0 s d þ c r; s 0 s 0; if W> a d þ c >: r: s 0 s The following theore gives the exected charge to a service request: Theore 2. The exected charge to a service request is P q C ¼ ar ððs rþðc=s 0 =sþþþ ; ððs rþða=d þ c=s 0 =sþþþ where P q ¼ =ð Þ and ¼ 0 ðþ =!. Sketch of the Proof. The roof is actually a detailed calculation of C, which contains two stes. In the first ste, we calculate CðrÞ, i.e., the exected charge to a service request with execution requireent r, based on the df of W obtained fro Theore. In the second ste, we calculate C based on the df of r. A colete roof of the theore is given in Section 9. tu In Fig., we consider the exected charge to a service request with execution requireent r, i.e., dp q CðrÞ ¼ar ð Þ e ð Þðc=s 0 =sþr e ð Þða=dþc=s0 =sþr : We assue that r ¼ billion instructions, ¼ 7 servers, s 0 ¼ billion instructions er second, s ¼ billion instructions er second, a ¼ 0 cents er one billion instructions (Note: The onetary unit cent in this aer ay not be identical but should be linearly roortional to the real cent in US dollars.), c ¼ 3, and d ¼ cents er second. (Note: These araeters are chosen only for illustration and should be scaled to any values.) For ¼ 6:5; 6:35; 6:55; 6:75; 6:95 service requests er second, we show CðrÞ for 0 r 3. It can be seen that the service charge is a decreasing function of, since the waiting tie and lateness enalty increase as increases. It can also be seen that the service charge is an increasing function of r, i.e., large service requests generate ore revenue than sall service requests. In Fig. 2, we further dislay CðrÞ=ar using the sae araeters in Fig.. Since ar is the ideal (axiu) charge to a service request with execution requireent r, CðrÞ=ar is considered as the noralized service charge. For ¼ 6:5; 6:35; 6:55; 6:75; 6:95 service requests er second, we show CðrÞ=ar for 0 r 3. It can be seen that the noralized service charge is a decreasing function of, since the waiting tie and lateness enalty increase as increases. It can also be seen that the noralized service charge is an increasing function of r, i.e., the ercentage of lost service charge due to waiting tie decreases as service requireent r increases. In other words, it is ore likely to ake rofit fro large service requests and it is ore likely to give free services to sall service requests. It can be verified that as r aroaches 0, the noralized service CðrÞ charge is li r!0 ar ¼ P q, where P q increases (and P q decreases) as increases. It can also be verified that as r aroaches infinity, the noralized service charge is CðrÞ li r! ar ¼, for all. 6 NET BUSINESS GAIN Since the nuber of service requests rocessed in one unit of tie is in a stable M/M/ queuing syste, the exected service charge in one unit of tie is C, which is actually the exected revenue of a service rovider. Assue that the rental cost of one server for unit of tie is. Also, assue that the cost of energy is er Watt. The cost of a service rovider is the su of the cost of infrastructure renting and the cost of energy consution, i.e., þ P. Then, the exected net business gain (i.e., the net rofit) of a service rovider in one unit of tie is G ¼ C ð þ PÞ, which is defined as the revenue inus the cost. The above equation is G ¼ C ð þ ðrs þ P ÞÞ, for the idle-seed odel, and G ¼ C ð þ ðs þ P ÞÞ, for the constant-seed odel. In Figs. 3 and 4, we deonstrate the revenue C and the net business gain G in one unit of tie as a function of for the two ower consution odels, resectively, using the sae araeters in Figs. and 2. Furtherore, we assue

6 092 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 24, NO. 6, JUNE 203 Fig. 3. Revenue and net business gain versus (idle-seed odel). that P ¼ 2 Watts, ¼ 2:0, ¼ 9:492, ¼ :5 cents er second, and ¼ 0: cents er Watt. For 0 7, we show C and G. The cost of infrastructure renting is ¼ 4 cents er second, and the cost of energy consution is 0:5 þ 7 cents er second for the idle-seed odel and 0.5 cents er second for the constant-seed odel. We observe that both C and G increase with alost linearly and dro sharly after certain oint. In other words, ore service requests bring ore revenue and net business gain; however, after the nuber of service requests er unit of tie reaches certain oint, the excessive waiting tie causes increased lateness enalty, so that there is no revenue and negative business gain. There are two situations that cause negative business gain. In the first case, there is no enough business (i.e., service requests). In this case, a service rovider should consider reducing the nuber of servers and/or server seed s, so that the cost of infrastructure renting and the cost of energy consution can be reduced. In the second case, there is too uch business (i.e., service requests). In this case, a service rovider should consider increasing the nuber of servers and/or server seed, so that the waiting tie can be reduced and the revenue can be increased. However, increasing the nuber of servers and/or server seed also increases the cost of infrastructure renting and the cost of energy consution. Therefore, we have the roble of selecting the otial server size and/or server seed so that the rofit is axiized. 7 PROFIT MAXIMIZATION To forulate and solve our otiization robles analytically, we need a closed-for exression of C. To this end, let P us use the following closed-for aroxiation, ðþ k k¼0 k! e, which is very accurate when is not too sall and is not too large [7]. We also need Stirling s aroxiation of!, i.e.,! ffiffiffiffiffiffiffiffiffi 2ð e Þ. Therefore, we get the following closed-for aroxiation of : ffiffiffiffiffiffiffiffiffi 2ð Þðe =eþ þ ; and the following closed-for aroxiation of P q : P q ffiffiffiffiffiffiffiffiffi 2ð Þðe =eþ þ : Fig. 4. Revenue and net business gain versus (constant-seed odel). By using the above-closed-for exression of P q, we get a closed-for aroxiation of the exected service charge to a service request as C ar ffiffiffiffiffiffiffiffiffi ð 2ð Þðe =eþ þ Þ ððs rþðc=s 0 =sþþþ! : ððs rþða=d þ c=s 0 =sþþþ For convenience, we rewrite C as C ¼ arð D D 2 D 3 Þ, where D ¼ ffiffiffiffiffiffiffiffiffi 2ð Þðe =eþ þ ; D 2 ¼ðs rþðc=s 0 =sþþ; D 3 ¼ðs rþða=d þ c=s 0 =sþþ: Our discussion in this section is based on the above-closedfor exression of C. 7. Otial Size Given, r, s, P,,,, a, c, and d, our first roble is to find such that G is axiized. To axiize G, we need to find @ ð þ P Þ¼0; for the idle-seed @ ð þ ðs þ P ÞÞ ¼ 0; for the constant-seed ¼ ar ðd D 2 D 3 Þ D 2 D þ 2 D þ 3 D 2 To continue the calculation, we rewrite D as D ¼ ffiffiffiffiffiffiffiffiffi 2ð ÞR þ, where R ¼ðe =eþ. Notice that ln R ¼ lnðe =eþ ¼ð ln Þ. r 2 s ¼ ;

7 CAO ET AL.: OPTIMAL MULTISERVER CONFIGURATION FOR PROFIT MAXIMIZATION IN CLOUD COMPUTING 093 Fig. 5. Net business gain G versus and (idle-seed odel). we ¼ð ln ¼ ln ¼ Rln Now, ¼ ffiffiffiffiffi 2 2 ffiffiffiffi ð ÞR þ þ ð ¼ ffiffiffiffiffi 2 2 ffiffiffiffi ð ÞR þ ffiffiffiffi R ffiffiffiffi ð ÞR ln ¼ ffiffiffiffiffi 2 2 ffiffiffiffi ð ÞR þ ffiffiffiffi R ffiffiffiffi ð Þðln ÞR ¼ ffiffiffiffiffi 2 2 ffiffiffiffi ð þ ÞR ffiffiffiffi ð Þðln ÞR : Furtherore, we have Fig. 6. Net business gain G versus and (constant-seed odel). Such server size otiization has clear hysical interretation. When is sall such that is close to, the waiting ties of service requests are excessively long, and the service charges and the net business gain are low. As increases, the waiting ties are significantly reduced, and the service charges and the net business gain are increased. However, as further increases, there will be no ore increase in the exected services charge which has an uer bound ar; on the other hand, the cost of a service rovider (i.e., the rental cost and base ower consution) increases, so that the net business gain is actually reduced. Hence, there is an otial choice of which axiizes the rofit. 7.2 Otial Seed Given, r,, P,,,, a, c, and d, our second roble is to find s such that G is axiized. To axiize G, we need to find s @s rð Þs 2 ¼ 0; for the idle-seed odel, and ¼ cs=s @s s ¼ 0; for the constant-seed odel, ¼ as=d þ cs=s 0 : Although there is no closed-for solution to, we notice is a decreasing function of. Therefore, can be found nuerically by using the standard bisection ethod. In Figs. 5 and 6, we deonstrate the net business gain G in one unit of tie as a function of and for the two ower consution odels, resectively, using the sae araeters in Figs., 2, 3, and 4. For ¼ 2:9; 3:9; 4:9; 5:9; 6:9, we dislay G for large enough such that <. We notice that there is an otial choice of such that G is axiized. Using our analytical results, we can find such ¼ 0. The otial value of is , , , , , resectively, for the idleseed odel, and , , , 6.860, , resectively, for the ¼ ar ðd D 2 D 3 Þ D 2 D þ 2 D þ 3 D 2 Siilar to the calculation in the last section, we r s ¼ 2 s ; @s ¼ ð Þ; ¼ ð ÞR: s

8 094 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 24, NO. 6, JUNE 203 Fig. 7. Net business gain G versus s and (idle-seed odel). Now, ¼ ffiffiffiffiffiffiffiffiffi R @s ¼ ffiffiffiffiffiffiffiffiffi 2 s R þ s ð Þ2 R ¼ ffiffiffiffiffiffiffiffiffi 2ð þ ð Þ 2 Þ R s : Furtherore, we ¼ c þðs rþ s 0 s s 2 ¼ c r s 0 s 2 ; ¼ a d þ c þðs rþ s 0 s s 2 ¼ a d þ c r s 0 s : 2 Although there is no closed-for solution to s, we notice is a decreasing function of s. Therefore, s can be found nuerically by using the standard bisection ethod. In Figs. 7 and 8, we deonstrate the net business gain G in one unit of tie as a function of s and for the two ower consution odels, resectively, using the sae araeters in Figs., 2, 3, 4, 5, and 6. For ¼ 2:9; 3:9; 4:9; 5:9; 6:9, we dislay G for s large enough such that <. We notice that there is an otial choice of s such that G is axiized. Using our analytical results, we can find s such that Fig. 9. Net business gain G versus s and (idle-seed ¼ 0. The otial value of s is , , ,.049,.90, resectively, for the idle-seed odel, and , , , ,.3584, resectively, for the constant-seed odel. Such server seed otiization also has clear hysical interretation. When s is sall such that is close to, the waiting ties of service requests are excessively long, and the service charges and the net business gain are low. As s increases, the waiting ties are significantly reduced, and the service charges and the net business gain are increased. However, as s further increases, there will be no ore increase in the exected services charge which has an uer bound ar; on the other hand, the cost of a service rovider (i.e., the cost of energy consution) increases, so that the net business gain is actually reduced. Hence, there is an otial choice of s which axiizes the rofit. 7.3 Otial Size and Seed Given, r, P,,,, a, c, and d, our third roble is to find and s such that G is axiized. To axiize G, we need to find and s such ¼ 0 ¼ 0, have been derived in the last two sections. The two equations can be solved by a nested bisection search rocedure. In Figs. 9 and 0, we deonstrate the net business gain G in one unit of tie as a function of s and for the two ower consution odels, resectively, using the sae araeters in Figs., 2, 3, 4, 5, 6, 7, and 8, where ¼ 6:9. For ¼ 4; 5; 6; 7; 8, we dislay G for s large enough such that <. Using our analytical results, we can find and s such ¼ 0 ¼ 0. For the idle-seed Fig. 8. Net business gain G versus s and (constant-seed odel). Fig. 0. Net business gain G versus s and (constant-seed odel).

9 CAO ET AL.: OPTIMAL MULTISERVER CONFIGURATION FOR PROFIT MAXIMIZATION IN CLOUD COMPUTING 095 odel, the theoretically otial values are ¼ 5:56827 and s ¼ :4689, which result in the axiu G ¼ 49:2536 by using the closed-for aroxiation of C. Practically, can be either 5 or 6. When ¼ 5, the otial value of s is.62236, which results in the axiu G ¼ 49:650. When ¼ 6, the otial value of s is.37044, which results in the axiu G ¼ 49:8888. Hence, the ractically otial setting is ¼ 6 and s ¼ :37044, and the axiu net business gain in one unit of tie is G ¼ 49:29273 by using the exact exression of C. For the constant-seed odel, the theoretically otial values are ¼ 5:79074 and s ¼ :35667, which result in the axiu G ¼ 47:80769 by using the closed-for aroxiation of C. Practically, can be either 5 or 6. When ¼ 5, the otial value of s is.55839, which results in the axiu G ¼ 47: When ¼ 6, the otial value of s is.323, which results in the axiu G ¼ 47: Hence, the ractically otial setting is ¼ 6 and s ¼ :323, and the axiu net business gain in one unit of tie is G ¼ 47:9830 by using the exact exression of C. 8 SIMULATION SETTINGS AND RESULTS See Section of the suleentary aterial, which can be found on the Couter Society Digital Library at htt:// doi.ieeecoutersociety.org/0.09/tpds PROOFS OF THEOREMS AND 2 See Section 2 of the suleentary aterial, available online. 0 CONCLUDING REMARKS We have roosed a ricing odel for cloud couting which takes any factors into considerations, such as the requireent r of a service, the workload of an alication environent, the configuration ( and s) of a ultiserver syste, the service level agreeent c, the satisfaction (r and s 0 ) of a consuer, the quality (W and T) of a service, the enalty d of a low-quality service, the cost ( and ) of renting, the cost (,, P, and P ) of energy consution, and a service rovider s argin and rofit a. By using an M/M/ queuing odel, we forulated and solved the roble of otial ultiserver configuration for rofit axiization in a cloud couting environent. Our discussion can be easily extended to other service charge functions. Our ethodology can be alied to other ricing odels. ACKNOWLEDGMENTS The authors are grateful to three anonyous reviewers for their constructive coents. Part of the work were erfored while K. Hwang, K. Li, and A.Y. Zoaya were visiting Tsinghua National Laboratory for Inforation Science and Technology, Tsinghua University, in the winter of 20 and the suer of 202 as Intellectual Ventures endowed visiting chair rofessors. This work is artially suorted by Ministry of Science and Technology of China under National 973 Basic Research Grants No. 20CB and No. 20CB302505, and National 863 High-tech Progra Grant No. 20AA04050; Ministry of Industry and Inforation Technology of China under the Internet of Things rogra; the Innovation R/D Tea Progra of Guangdong Province, China, under contract No. 2000D ; Australian Research Grant DP0970. REFERENCES [] htt://en.wikiedia.org/wiki/cmos, 202. [2] htt://en.wikiedia.org/wiki/service_level_agreeent, 202. [3] M. Arbrust et al., Above the Clouds: A Berkeley View of Cloud Couting, Technical Reort No. UCB/EECS , Feb [4] R. Buyya, D. Abrason, J. Giddy, and H. Stockinger, Econoic Models for Resource Manageent and Scheduling in Grid Couting, Concurrency and Coutation: Practice and Exerience, vol. 4, , [5] R. Buyya, C.S. Yeo, S. Venugoal, J. Broberg, and I. Brandic, Cloud Couting and Eerging IT Platfors: Vision, Hye, and Reality for Delivering Couting as the Fifth Utility, Future Generation Couter Systes, vol. 25, no. 6, , [6] A.P. Chandrakasan, S. Sheng, and R.W. Brodersen, Low-Power CMOS Digital Design, IEEE J. Solid-State Circuits, vol. 27, no. 4, , Ar [7] B.N. Chun and D.E. Culler, User-Centric Perforance Analysis of Market-Based Cluster Batch Schedulers, Proc. Second IEEE/ ACM Int l Sy. Cluster Couting and the Grid, [8] D. Durkee, Why Cloud Couting Will Never be Free, Co. ACM, vol. 53, no. 5, , 200. [9] R. Ghosh, K.S. Trivedi, V.K. Naik, and D.S. Ki, End-to-End Perforability Analysis for Infrastructure-as-a-Service Cloud: An Interacting Stochastic Models Aroach, Proc. 6th IEEE Pacific Ri Int l Sy. Deendable Couting, , 200. [0] K. Hwang, G.C. Fox, and J.J. Dongarra, Distributed and Cloud Couting. Morgan Kaufann, 202. [] Enhanced Intel SeedSte Technology for the Intel Pentiu M Processor, White Paer, Intel, Mar [2] D.E. Irwin, L.E. Grit, and J.S. Chase, Balancing Risk and Reward in a Market-Based Task Service, Proc. 3th IEEE Int l Sy. High Perforance Distributed Couting, , [3] H. Khazaei, J. Misic, and V.B. Misic, Perforance Analysis of Cloud Couting Centers Using M/G//+r Queuing Systes, IEEE Trans. Parallel and Distributed Systes, vol. 23, no. 5, , May 202. [4] L. Kleinrock, Queueing Systes: Theory, vol.. John Wiley and Sons, 975. [5] Y.C. Lee, C. Wang, A.Y. Zoaya, and B.B. Zhou, Profit-Driven Service Request Scheduling in Clouds, Proc. 0th IEEE/ACM Int l Conf. Cluster, Cloud and Grid Couting,. 5-24, 200. [6] K. Li, Otial Load Distribution for Multile Heterogeneous Blade Servers in a Cloud Couting Environent, Proc. 25th IEEE Int l Parallel and Distributed Processing Sy. Workshos, , May 20. [7] K. Li, Otial Configuration of a Multicore Server Processor for Managing the Power and Perforance Tradeoff, J. Suercouting, vol. 6, no., , 202. [8] P. Mell and T. Grance, The NIST Definition of Cloud Couting, Nat l Inst. of Standards and Technology, htt://csrc.nist. gov/grous/sns/cloud-couting/, [9] F.I. Poovici and J. Wilkes, Profitable Services in an Uncertain World, Proc. ACM/IEEE Conf. Suercouting, [20] J. Sherwani, N. Ali, N. Lotia, Z. Hayat, and R. Buyya, Libra: A Coutational Econoy-Based Job Scheduling Syste for Clusters, Software - Practice and Exerience, vol. 34, , [2] C.S. Yeo and R. Buyya, A Taxonoy of Market-Based Resource Manageent Systes for Utility-Driven Cluster Couting, Software - Practice and Exerience, vol. 36, , [22] B. Zhai, D. Blaauw, D. Sylvester, and K. Flautner, Theoretical and Practical Liits of Dynaic Voltage Scaling, Proc. 4st Design Autoation Conf., , 2004.

10 096 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 24, NO. 6, JUNE 203 Junwei Cao received the bachelor s and aster s degrees in control theories and engineering in 996 and 998, resectively, both fro Tsinghua University, Beijing, China. He received the PhD degree in couter science fro the University of Warwick, Coventry, United Kingdo, in 200. He is currently a rofessor and vice director, Research Institute of Inforation Technology, Tsinghua University, Beijing, China. He is also the director of Coon Platfor and Technology Division, Tsinghua National Laboratory for Inforation Science and Technology. Before joining Tsinghua University in 2006, he was a research scientist at MIT LIGO Laboratory and NEC Laboratories Euroe for about five years. He has ublished ore than 30 aers and cited by international scholars for over 3,000 ties. He is the book editor of Cyberinfrastructure Technologies and Alications, ublished by Nova Science in His research is focused on advanced couting technologies and alications. He is a senior eber of the IEEE and IEEE Couter Society, and a eber of the ACM and CCF. Kai Hwang received the PhD degree fro the University of California, Berkeley in 972. He is a rofessor of EE/CS at the University of Southern California. He also chairs the IV-endowed visiting chair rofessor grou at Tsinghua University in China. He has ublished eight books and ore than 28 scientific aers in couter architecture, arallel rocessing, distributed systes, cloud couting, network security, and Internet alications. His oular books have been adoted worldwide and translated into four foreign languages. His ublished aers have been cited ore than,000 ties by early 202. His latest book Distributed and Cloud Couting: fro Parallel Processing to the Internet of Things (with G. Fox and J. Dongarra) was just ublished by Kaufann in 20. He received the 2004 CFC Outstanding Achieveent Award, and the Founders Award for his ioneering work in arallel rocessing fro IEEE IPDPS in 20. He has served as a founding editor-in-chief of the Journal of Parallel and Distributed Couting for 28 years. He has delivered 34 keynote addresses on advanced couting systes and cutting-edge inforation technologies in ajor IEEE/ACM Conferences. He has erfored advisory, consulting and collaborative work for IBM, Intel, MIT Lincoln Lab, JPL at Caltech, ETL in Jaan, ITRI in Taiwan, GMD in Gerany, INRIA in France, and Chinese Acadey of Sciences. He is a fellow of the IEEE (986). Keqin Li is a SUNY distinguished rofessor in couter science and an Intellectual Ventures endowed visiting chair rofessor at Tsinghua University, China. His research interests are ainly in design and analysis of algoriths, arallel and distributed couting, and couter networking. He has contributed extensively to rocessor allocation and resource anageent; design and analysis of sequential/arallel, deterinistic/robabilistic, and aroxiation algoriths; arallel and distributed couting systes erforance analysis, rediction, and evaluation; job scheduling, task disatching, and load balancing in heterogeneous distributed systes; dynaic tree ebedding and randoized load distribution in static networks; arallel couting using otical interconnections; dynaic location anageent in wireless counication networks; routing and wavelength assignent in otical networks; energy-efficient ower anageent and erforance otiization. He has ublished ore than 240 research ublications and has received several Best Paer Awards for his highest quality work. He is currently on the editorial board of IEEE Transactions on Parallel and Distributed Systes, IEEE Transactions on Couters, International Journal of Parallel, Eergent and Distributed Systes, International Journal of High Perforance Couting and Networking, and Otiization Letters. He is a senior eber of the IEEE. Albert Y. Zoaya is currently the chair rofessor of high erforance couting and networking and Australian Research Council Professorial fellow in the School of Inforation Technologies, The University of Sydney. He is also the director of the Centre for Distributed and High Perforance Couting which was established in late He is the author/coauthor of seven books, ore than 400 aers, and the editor of nine books and conference roceedings. He is the editor-in-chief of the IEEE Transactions on Couters and serves as an associate editor for 9 leading journals, such as, the IEEE Transactions on Parallel and Distributed Systes and Journal of Parallel and Distributed Couting. He is the reciient of the Meritorious Service Award (in 2000) and the Golden Core Recognition (in 2006), both fro the IEEE Couter Society. Also, he received the IEEE Technical Coittee on Parallel Processing Outstanding Service Award and the IEEE Technical Coittee on Scalable Couting Medal for Excellence in Scalable Couting, both in 20. He is a chartered engineer, a fellow of AAAS, IEEE, and IET (United Kingdo).. For ore inforation on this or any other couting toic, lease visit our Digital Library at

IJESRT. Scientific Journal Impact Factor: (ISRA), Impact Factor: 1.852

IJESRT. Scientific Journal Impact Factor: (ISRA), Impact Factor: 1.852 IJESRT INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH TECHNOLOGY Dynamic Resource Allocation Using VM for Cloud Computing Environment V.Indhumathi *1, N.Nagalakshmi 2 *1, 2 Student, Department

More information

Optimization of Multi-server Configuration for Profit Maximization using M/M/m Queuing Model

Optimization of Multi-server Configuration for Profit Maximization using M/M/m Queuing Model International Journal of Computer Sciences and Engineering Open Access Research Paper Volume-2, Issue-8 E-ISSN: 2347-2693 Optimization of Multi-server Configuration for Profit Maximization using M/M/m

More information

A Fail-Aware Datagram Service

A Fail-Aware Datagram Service A Fail-Aware Datagra Service Christof Fetzer and Flaviu Cristian christof@research.att.co, htt://www.christof.org Abstract In distributed real-tie systes it is often useful for a rocess to know that another

More information

EXTENDED SVD FLATNESS CONTROL. Per Erik Modén and Markus Holm ABB AB, Västerås, Sweden

EXTENDED SVD FLATNESS CONTROL. Per Erik Modén and Markus Holm ABB AB, Västerås, Sweden EXTENDED SVD FLATNESS CONTROL Per Erik Modén and Markus Hol ABB AB, Västerås, Sweden ABSTRACT Cold rolling ills soeties do not see able to control flatness as well as exected, taking into account the nuber

More information

A Fail-Aware Datagram Service

A Fail-Aware Datagram Service A Fail-Aware Datagra Service Christof Fetzer and Flaviu Cristian christof@research.att.co, htt://www.christof.org Abstract In distributed real-tie systes it is often useful for a rocess Ô to know that

More information

I n many cases, the SPRT will come to a decision with fewer samples than would have been required for a fixed size test.

I n many cases, the SPRT will come to a decision with fewer samples than would have been required for a fixed size test. STATGRAPHICS Rev. 9/6/3 Sequential Saling Suary... Data Inut... 3 Analysis Otions... 3 Analysis Suary... 5 Cuulative Plot... 6 Decision Nubers... 9 Test Perforance... O. C. Curve... ASN Function... Forulas...

More information

On-Chip Interconnect Implications of Shared Memory Multicores

On-Chip Interconnect Implications of Shared Memory Multicores On-Chi Interconnect Ilications of Shared Meory Multicores Srini Devadas Couter Science and Artificial Intelligence Laboratory (CSAIL) Massachusetts Institute of Technology 1 Prograing 1000 cores MPI has

More information

New method of angle error measurement in angular artifacts using minimum zone flatness plane

New method of angle error measurement in angular artifacts using minimum zone flatness plane Alied Mechanics and Materials Subitted: 04-05-4 ISSN: 66-748, Vols. 599-60, 997-004 Acceted: 04-06-05 doi:0.408/www.scientific.net/amm.599-60.997 Online: 04-08- 04 Trans Tech Publications, Switzerland

More information

Markov Analysis for Optimum Caching as an Alternative to Belady s Algorithm

Markov Analysis for Optimum Caching as an Alternative to Belady s Algorithm arov Analysis for Otiu Caching as an Alternative to Belady s Algorith, Deutsche Teleo, Darstadt, Gerany gerhard.hasslinger@teleo.de Analytic Results on LRU, LFU, Otiu Caching Belady s Princile for Otiu

More information

Recap Consistent cuts. CS514: Intermediate Course in Operating Systems. What time is it? But what does time mean? Drawing time-line pictures:

Recap Consistent cuts. CS514: Intermediate Course in Operating Systems. What time is it? But what does time mean? Drawing time-line pictures: CS514: Interediate Course in Oerating Systes Professor Ken iran Vivek Vishnuurthy: T Reca Consistent cuts On Monday we saw that sily gathering the state of a syste isn t enough Often the state includes

More information

Performance Analysis of RAID in Different Workload

Performance Analysis of RAID in Different Workload Send Orders for Reprints to reprints@benthascience.ae 324 The Open Cybernetics & Systeics Journal, 2015, 9, 324-328 Perforance Analysis of RAID in Different Workload Open Access Zhang Dule *, Ji Xiaoyun,

More information

PERFORMANCE MEASURES FOR INTERNET SERVER BY USING M/M/m QUEUEING MODEL

PERFORMANCE MEASURES FOR INTERNET SERVER BY USING M/M/m QUEUEING MODEL IJRET: International Journal of Research in Engineering and Technology ISSN: 239-63 PERFORMANCE MEASURES FOR INTERNET SERVER BY USING M/M/ QUEUEING MODEL Raghunath Y. T. N. V, A. S. Sravani 2 Assistant

More information

Derivation of an Analytical Model for Evaluating the Performance of a Multi- Queue Nodes Network Router

Derivation of an Analytical Model for Evaluating the Performance of a Multi- Queue Nodes Network Router Derivation of an Analytical Model for Evaluating the Perforance of a Multi- Queue Nodes Network Router 1 Hussein Al-Bahadili, 1 Jafar Ababneh, and 2 Fadi Thabtah 1 Coputer Inforation Systes Faculty of

More information

An energy-efficient random verification protocol for the detection of node clone attacks in wireless sensor networks

An energy-efficient random verification protocol for the detection of node clone attacks in wireless sensor networks Zhou et al. EURASIP Journal on Wireless Counications and Networking 2014, 2014:163 htt://jwcn.eurasijournals.co/content/2014/1/163 RESEARCH Oen Access An energy-efficient rando verification rotocol for

More information

Utility-Based Resource Allocation for Mixed Traffic in Wireless Networks

Utility-Based Resource Allocation for Mixed Traffic in Wireless Networks IEEE IFOCO 2 International Workshop on Future edia etworks and IP-based TV Utility-Based Resource Allocation for ixed Traffic in Wireless etworks Li Chen, Bin Wang, Xiaohang Chen, Xin Zhang, and Dacheng

More information

The optimization design of microphone array layout for wideband noise sources

The optimization design of microphone array layout for wideband noise sources PROCEEDINGS of the 22 nd International Congress on Acoustics Acoustic Array Systes: Paper ICA2016-903 The optiization design of icrophone array layout for wideband noise sources Pengxiao Teng (a), Jun

More information

Interactive Fuzzy Modeling by Evolutionary Multiobjective Optimization with User Preference

Interactive Fuzzy Modeling by Evolutionary Multiobjective Optimization with User Preference Interactive Fuzzy Modeling by Evolutionary Multiobjective Otiization with User Preference Yusue ojia, Hisao Ishibuchi Deartent of Couter Science and Intelligent Systes, Osaa Prefecture University - Gauen-cho,

More information

Investigation of The Time-Offset-Based QoS Support with Optical Burst Switching in WDM Networks

Investigation of The Time-Offset-Based QoS Support with Optical Burst Switching in WDM Networks Investigation of The Tie-Offset-Based QoS Support with Optical Burst Switching in WDM Networks Pingyi Fan, Chongxi Feng,Yichao Wang, Ning Ge State Key Laboratory on Microwave and Digital Counications,

More information

Different criteria of dynamic routing

Different criteria of dynamic routing Procedia Coputer Science Volue 66, 2015, Pages 166 173 YSC 2015. 4th International Young Scientists Conference on Coputational Science Different criteria of dynaic routing Kurochkin 1*, Grinberg 1 1 Kharkevich

More information

MAC schemes - Fixed-assignment schemes

MAC schemes - Fixed-assignment schemes MAC schees - Fixed-assignent schees M. Veeraraghavan, April 6, 04 Mediu Access Control (MAC) schees are echaniss for sharing a single link. MAC schees are essentially ultiplexing schees. For exaple, on

More information

Performance analysis of hybrid (M/M/1 and M/M/m) client server model using Queuing theory

Performance analysis of hybrid (M/M/1 and M/M/m) client server model using Queuing theory International Journal of Electronic and Couter cience Engineering vailable Online at wwwijeceorg IN- 77-9 erforance analyi of hybrid M/M/ and M/M/ client erver odel uing ueuing theory atarhi Guta, Dr Rajan

More information

Energy-Efficient Disk Replacement and File Placement Techniques for Mobile Systems with Hard Disks

Energy-Efficient Disk Replacement and File Placement Techniques for Mobile Systems with Hard Disks Energy-Efficient Disk Replaceent and File Placeent Techniques for Mobile Systes with Hard Disks Young-Jin Ki School of Coputer Science & Engineering Seoul National University Seoul 151-742, KOREA youngjk@davinci.snu.ac.kr

More information

Database Design on Mechanical Equipment Operation Management System Zheng Qiu1, Wu kaiyuan1, Wu Chengyan1, Liu Lei2

Database Design on Mechanical Equipment Operation Management System Zheng Qiu1, Wu kaiyuan1, Wu Chengyan1, Liu Lei2 2nd International Conference on Advances in Mechanical Engineering and Industrial Inforatics (AMEII 206) Database Design on Mechanical Equipent Manageent Syste Zheng Qiu, Wu kaiyuan, Wu Chengyan, Liu Lei2

More information

The Application of Bandwidth Optimization Technique in SLA Negotiation Process

The Application of Bandwidth Optimization Technique in SLA Negotiation Process The Application of Bandwidth Optiization Technique in SLA egotiation Process Srećko Krile University of Dubrovnik Departent of Electrical Engineering and Coputing Cira Carica 4, 20000 Dubrovnik, Croatia

More information

Evaluation of a multi-frame blind deconvolution algorithm using Cramér-Rao bounds

Evaluation of a multi-frame blind deconvolution algorithm using Cramér-Rao bounds Evaluation of a ulti-frae blind deconvolution algorith using Craér-Rao bounds Charles C. Beckner, Jr. Air Force Research Laboratory, 3550 Aberdeen Ave SE, Kirtland AFB, New Mexico, USA 87117-5776 Charles

More information

A Novel Architecture for Compiled-type Software CNC System

A Novel Architecture for Compiled-type Software CNC System Key Engineering Materials Online: 2007-05-15 ISSN: 1662-9795, ol. 339, 442-446 doi:10.4028/.scientific.net/kem.339.442 2007 rans ech Pulications, Sitzerland A Novel Architecture for Coiled-tye Softare

More information

The Internal Conflict of a Belief Function

The Internal Conflict of a Belief Function The Internal Conflict of a Belief Function Johan Schubert Abstract In this paper we define and derive an internal conflict of a belief function We decopose the belief function in question into a set of

More information

A Distributed Multi-service Resource Allocation Algorithm in Heterogeneous Wireless Access Medium

A Distributed Multi-service Resource Allocation Algorithm in Heterogeneous Wireless Access Medium 1 A Distributed Multi-service Resource Allocation Algorith in Heterogeneous Wireless Access Mediu Muhaad Isail, Student Meber, IEEE, and Weihua Zhuang, Fellow, IEEE Abstract In this paper, radio resource

More information

Measuring Bottleneck Bandwidth of Targeted Path Segments

Measuring Bottleneck Bandwidth of Targeted Path Segments Measuring Bottleneck Bandwidth of Targeted Path Segents Khaled Harfoush Deartent of Couter Science North Carolina State University Raleigh, NC 27695 harfoush@cs.ncsu.edu Azer Bestavros Couter Science Deartent

More information

Computer Aided Drafting, Design and Manufacturing Volume 26, Number 2, June 2016, Page 13

Computer Aided Drafting, Design and Manufacturing Volume 26, Number 2, June 2016, Page 13 Coputer Aided Drafting, Design and Manufacturing Volue 26, uber 2, June 2016, Page 13 CADDM 3D reconstruction of coplex curved objects fro line drawings Sun Yanling, Dong Lijun Institute of Mechanical

More information

OPTIMAL COMPLEX SERVICES COMPOSITION IN SOA SYSTEMS

OPTIMAL COMPLEX SERVICES COMPOSITION IN SOA SYSTEMS Key words SOA, optial, coplex service, coposition, Quality of Service Piotr RYGIELSKI*, Paweł ŚWIĄTEK* OPTIMAL COMPLEX SERVICES COMPOSITION IN SOA SYSTEMS One of the ost iportant tasks in service oriented

More information

Knowledge Discovery Applied to Agriculture Economy Planning

Knowledge Discovery Applied to Agriculture Economy Planning Knowledge Discovery Applied to Agriculture Econoy Planning Bing-ru Yang and Shao-un Huang Inforation Engineering School University of Science and Technology, Beiing, China, 100083 Eail: bingru.yang@b.col.co.cn

More information

Scheduling Parallel Real-Time Recurrent Tasks on Multicore Platforms

Scheduling Parallel Real-Time Recurrent Tasks on Multicore Platforms IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL., NO., NOV 27 Scheduling Parallel Real-Tie Recurrent Tasks on Multicore Platfors Risat Pathan, Petros Voudouris, and Per Stenströ Abstract We

More information

Modeling Parallel Applications Performance on Heterogeneous Systems

Modeling Parallel Applications Performance on Heterogeneous Systems Modeling Parallel Applications Perforance on Heterogeneous Systes Jaeela Al-Jaroodi, Nader Mohaed, Hong Jiang and David Swanson Departent of Coputer Science and Engineering University of Nebraska Lincoln

More information

Analysis of a Biologically-Inspired System for Real-time Object Recognition

Analysis of a Biologically-Inspired System for Real-time Object Recognition Cognitive Science Online, Vol.3.,.-4, 5 htt://cogsci-online.ucsd.edu Analysis of a Biologically-Insired Syste for Real-tie Object Recognition Erik Murhy-Chutorian,*, Sarah Aboutalib & Jochen Triesch,3

More information

A Periodic Dynamic Load Balancing Method

A Periodic Dynamic Load Balancing Method 2016 3 rd International Conference on Engineering Technology and Application (ICETA 2016) ISBN: 978-1-60595-383-0 A Periodic Dynaic Load Balancing Method Taotao Fan* State Key Laboratory of Inforation

More information

Performance Analysis and QoE-aware Enhancement for IEEE e EDCA under Unsaturations

Performance Analysis and QoE-aware Enhancement for IEEE e EDCA under Unsaturations Perforance Analysis and QoE-aware Enhanceent for IEEE 80.11e EDCA under Unsaturations Yueying Zhang, Wei Li, Junlong Zhang, Hang Long, Wenbo Wang Wireless Signal Processing and etwork Lab Key laboratory

More information

Gromov-Hausdorff Distance Between Metric Graphs

Gromov-Hausdorff Distance Between Metric Graphs Groov-Hausdorff Distance Between Metric Graphs Jiwon Choi St Mark s School January, 019 Abstract In this paper we study the Groov-Hausdorff distance between two etric graphs We copute the precise value

More information

Redundancy Level Impact of the Mean Time to Failure on Wireless Sensor Network

Redundancy Level Impact of the Mean Time to Failure on Wireless Sensor Network (IJACSA) International Journal of Advanced Coputer Science and Applications Vol. 8, No. 1, 217 Redundancy Level Ipact of the Mean Tie to Failure on Wireless Sensor Network Alaa E. S. Ahed 1 College of

More information

Shortest Path Determination in a Wireless Packet Switch Network System in University of Calabar Using a Modified Dijkstra s Algorithm

Shortest Path Determination in a Wireless Packet Switch Network System in University of Calabar Using a Modified Dijkstra s Algorithm International Journal of Engineering and Technical Research (IJETR) ISSN: 31-869 (O) 454-4698 (P), Volue-5, Issue-1, May 16 Shortest Path Deterination in a Wireless Packet Switch Network Syste in University

More information

International Journal of Scientific & Engineering Research, Volume 6, Issue 9, September ISSN

International Journal of Scientific & Engineering Research, Volume 6, Issue 9, September ISSN International Journal of Scientific & Engineering Research, Volue 6, Issue 9, Seteber-05 9 ISSN 9-558 Doinating Sets Of Divisor Cayley Grahs G. Mirona, Prof. B. Maheswari Abstract Let n be an integer and

More information

Analysing Real-Time Communications: Controller Area Network (CAN) *

Analysing Real-Time Communications: Controller Area Network (CAN) * Analysing Real-Tie Counications: Controller Area Network (CAN) * Abstract The increasing use of counication networks in tie critical applications presents engineers with fundaental probles with the deterination

More information

1 Extended Boolean Model

1 Extended Boolean Model 1 EXTENDED BOOLEAN MODEL It has been well-known that the Boolean odel is too inflexible, requiring skilful use of Boolean operators to obtain good results. On the other hand, the vector space odel is flexible

More information

A Novel Fast Constructive Algorithm for Neural Classifier

A Novel Fast Constructive Algorithm for Neural Classifier A Novel Fast Constructive Algorith for Neural Classifier Xudong Jiang Centre for Signal Processing, School of Electrical and Electronic Engineering Nanyang Technological University Nanyang Avenue, Singapore

More information

QUERY ROUTING OPTIMIZATION IN SENSOR COMMUNICATION NETWORKS

QUERY ROUTING OPTIMIZATION IN SENSOR COMMUNICATION NETWORKS QUERY ROUTING OPTIMIZATION IN SENSOR COMMUNICATION NETWORKS Guofei Jiang and George Cybenko Institute for Security Technology Studies and Thayer School of Engineering Dartouth College, Hanover NH 03755

More information

An Ensemble of Adaptive Neuro-Fuzzy Kohonen Networks for Online Data Stream Fuzzy Clustering

An Ensemble of Adaptive Neuro-Fuzzy Kohonen Networks for Online Data Stream Fuzzy Clustering An Enseble of Adative euro-fuzzy Kohonen etworks for Online Data Strea Fuzzy Clustering Zhengbing Hu School of Educational Inforation Technology Central China oral University Wuhan China Eail: hzb@ail.ccnu.edu.cn

More information

ACNS: Adaptive Complementary Neighbor Selection in BitTorrent-like Applications

ACNS: Adaptive Complementary Neighbor Selection in BitTorrent-like Applications ACNS: Adaptive Copleentary Neighbor Selection in BitTorrent-like Applications Zhenbao Zhou 1, 2, Zhenyu Li 1, 2, Gaogang Xie 1 1 Institute of Coputing Technology, Chinese Acadey of Sciences, Beijing 100190,

More information

COLOR HISTOGRAM AND DISCRETE COSINE TRANSFORM FOR COLOR IMAGE RETRIEVAL

COLOR HISTOGRAM AND DISCRETE COSINE TRANSFORM FOR COLOR IMAGE RETRIEVAL COLOR HISTOGRAM AND DISCRETE COSINE TRANSFORM FOR COLOR IMAGE RETRIEVAL 1 Te-Wei Chiang ( 蔣德威 ), 2 Tienwei Tsai ( 蔡殿偉 ), 3 Jeng-Ping Lin ( 林正平 ) 1 Dept. of Accounting Inforation Systes, Chilee Institute

More information

A Low-Cost Multi-Failure Resilient Replication Scheme for High Data Availability in Cloud Storage

A Low-Cost Multi-Failure Resilient Replication Scheme for High Data Availability in Cloud Storage 216 IEEE 23rd International Conference on High Perforance Coputing A Low-Cost Multi-Failure Resilient Replication Schee for High Data Availability in Cloud Storage Jinwei Liu* and Haiying Shen *Departent

More information

An Efficient Approach for Content Delivery in Overlay Networks

An Efficient Approach for Content Delivery in Overlay Networks An Efficient Approach for Content Delivery in Overlay Networks Mohaad Malli, Chadi Barakat, Walid Dabbous Projet Planète, INRIA-Sophia Antipolis, France E-ail:{alli, cbarakat, dabbous}@sophia.inria.fr

More information

Closing The Performance Gap between Causal Consistency and Eventual Consistency

Closing The Performance Gap between Causal Consistency and Eventual Consistency Closing The Perforance Gap between Causal Consistency and Eventual Consistency Jiaqing Du Călin Iorgulescu Aitabha Roy Willy Zwaenepoel EPFL ABSTRACT It is well known that causal consistency is ore expensive

More information

An Efficient Video Program Delivery algorithm in Tree Networks*

An Efficient Video Program Delivery algorithm in Tree Networks* 3rd International Symosium on Parallel Architectures, Algorithms and Programming An Efficient Video Program Delivery algorithm in Tree Networks* Fenghang Yin 1 Hong Shen 1,2,** 1 Deartment of Comuter Science,

More information

Data & Knowledge Engineering

Data & Knowledge Engineering Data & Knowledge Engineering 7 (211) 17 187 Contents lists available at ScienceDirect Data & Knowledge Engineering journal hoepage: www.elsevier.co/locate/datak An approxiate duplicate eliination in RFID

More information

Author. Published. Journal Title DOI. Copyright Statement. Downloaded from. Griffith Research Online. Kandjani, Hadi, Wen, Larry, Bernus, Peter

Author. Published. Journal Title DOI. Copyright Statement. Downloaded from. Griffith Research Online. Kandjani, Hadi, Wen, Larry, Bernus, Peter Enterprise Architecture Cybernetics for Global Mining Projects: Reducing the Structural Coplexity of Global Mining Supply Networks via Virtual Brokerage Author Kandjani, Hadi, Wen, Larry, Bernus, Peter

More information

A Preliminary Study of Maximum System-level Crosstalk at High Frequencies for Coupled Transmission Lines

A Preliminary Study of Maximum System-level Crosstalk at High Frequencies for Coupled Transmission Lines A Preliinary tudy of Maxiu yste-leel Crosstalk at High Frequencies for Couled Transission Lines iaoeng Dong, Haixiao Weng, Daryl G Beetner Todd Hubing Electroagnetic Coatibility Laboratory Uniersity of

More information

Fair Energy-Efficient Sensing Task Allocation in Participatory Sensing with Smartphones

Fair Energy-Efficient Sensing Task Allocation in Participatory Sensing with Smartphones Advance Access publication on 24 February 2017 The British Coputer Society 2017. All rights reserved. For Perissions, please eail: journals.perissions@oup.co doi:10.1093/cojnl/bxx015 Fair Energy-Efficient

More information

Identifying Converging Pairs of Nodes on a Budget

Identifying Converging Pairs of Nodes on a Budget Identifying Converging Pairs of Nodes on a Budget Konstantina Lazaridou Departent of Inforatics Aristotle University, Thessaloniki, Greece konlaznik@csd.auth.gr Evaggelia Pitoura Coputer Science and Engineering

More information

Spatial interference encoding patterns based super resolved photoacoustic microscopy

Spatial interference encoding patterns based super resolved photoacoustic microscopy Satial interference encoding atterns based suer resolved hotoacoustic icroscoy Aihai Meiri 1, Eric M. Stroh 2, Michael C. Kolios 2 and Zeev Zalevsky 1 1 Faculty of Engineering and the Nano Technology Center,

More information

A Hybrid Network Architecture for File Transfers

A Hybrid Network Architecture for File Transfers JOURNAL OF IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 6, NO., JANUARY 9 A Hybrid Network Architecture for File Transfers Xiuduan Fang, Meber, IEEE, Malathi Veeraraghavan, Senior Meber,

More information

Efficient Estimation of Inclusion Coefficient using HyperLogLog Sketches

Efficient Estimation of Inclusion Coefficient using HyperLogLog Sketches Efficient Estiation of Inclusion Coefficient using HyperLogLog Sketches Azade Nazi, Bolin Ding, Vivek Narasayya, Surajit Chaudhuri Microsoft Research {aznazi, bolind, viveknar, surajitc}@icrosoft.co ABSTRACT

More information

Secure Wireless Multihop Transmissions by Intentional Collisions with Noise Wireless Signals

Secure Wireless Multihop Transmissions by Intentional Collisions with Noise Wireless Signals Int'l Conf. Wireless etworks ICW'16 51 Secure Wireless Multihop Transissions by Intentional Collisions with oise Wireless Signals Isau Shiada 1 and Hiroaki Higaki 1 1 Tokyo Denki University, Japan Abstract

More information

TOLLY. Premise: Buyers of multi-port. BayStack T Switch Fast Ethernet Switch Competitive Evaluation Backplane Capacity.

TOLLY. Premise: Buyers of multi-port. BayStack T Switch Fast Ethernet Switch Competitive Evaluation Backplane Capacity. Preise: Buyers of ulti-port switches need to know the actual capacity of the backplane (switching fabric) used to interconnect ports. To assure a non-blocking environent i.e., all ports can run at wire-speed

More information

Randomized algorithms: Two examples and Yao s Minimax Principle

Randomized algorithms: Two examples and Yao s Minimax Principle Randomized algorithms: Two examles and Yao s Minimax Princile Maximum Satisfiability Consider the roblem Maximum Satisfiability (MAX-SAT). Bring your knowledge u-to-date on the Satisfiability roblem. Maximum

More information

S16-02, URL:

S16-02, URL: Self Introduction A/Prof ay Seng Chuan el: Email: scitaysc@nus.edu.sg Office: S-0, Dean s s Office at Level URL: htt://www.hysics.nus.edu.sg/~hytaysc I was a rogrammer from to. I have been working in NUS

More information

Deterministic Voting in Distributed Systems Using Error-Correcting Codes

Deterministic Voting in Distributed Systems Using Error-Correcting Codes IEEE TRASACTIOS O PARALLEL AD DISTRIBUTED SYSTEMS, VOL. 9, O. 8, AUGUST 1998 813 Deterinistic Voting in Distributed Systes Using Error-Correcting Codes Lihao Xu and Jehoshua Bruck, Senior Meber, IEEE Abstract

More information

Mapping Data in Peer-to-Peer Systems: Semantics and Algorithmic Issues

Mapping Data in Peer-to-Peer Systems: Semantics and Algorithmic Issues Mapping Data in Peer-to-Peer Systes: Seantics and Algorithic Issues Anastasios Keentsietsidis Marcelo Arenas Renée J. Miller Departent of Coputer Science University of Toronto {tasos,arenas,iller}@cs.toronto.edu

More information

Relocation of Gateway for Enhanced Timeliness in Wireless Sensor Networks Abstract 1. Introduction

Relocation of Gateway for Enhanced Timeliness in Wireless Sensor Networks Abstract 1. Introduction Relocation of ateway for Enhanced Tieliness in Wireless Sensor Networks Keal Akkaya and Mohaed Younis Departent of oputer Science and Electrical Engineering University of Maryland, altiore ounty altiore,

More information

Sensitivity Analysis for an Optimal Routing Policy in an Ad Hoc Wireless Network

Sensitivity Analysis for an Optimal Routing Policy in an Ad Hoc Wireless Network 1 Sensitivity Analysis for an Otimal Routing Policy in an Ad Hoc Wireless Network Tara Javidi and Demosthenis Teneketzis Deartment of Electrical Engineering and Comuter Science University of Michigan Ann

More information

A CRYPTANALYTIC ATTACK ON RC4 STREAM CIPHER

A CRYPTANALYTIC ATTACK ON RC4 STREAM CIPHER A CRYPTANALYTIC ATTACK ON RC4 STREAM CIPHER VIOLETA TOMAŠEVIĆ, SLOBODAN BOJANIĆ 2 and OCTAVIO NIETO-TALADRIZ 2 The Mihajlo Pupin Institute, Volgina 5, 000 Belgrade, SERBIA AND MONTENEGRO 2 Technical University

More information

RECONFIGURABLE AND MODULAR BASED SYNTHESIS OF CYCLIC DSP DATA FLOW GRAPHS

RECONFIGURABLE AND MODULAR BASED SYNTHESIS OF CYCLIC DSP DATA FLOW GRAPHS RECONFIGURABLE AND MODULAR BASED SYNTHESIS OF CYCLIC DSP DATA FLOW GRAPHS AWNI ITRADAT Assistant Professor, Departent of Coputer Engineering, Faculty of Engineering, Hasheite University, P.O. Box 15459,

More information

A GPU Heterogeneous Cluster Scheduling Model for Preventing Temperature Heat Island

A GPU Heterogeneous Cluster Scheduling Model for Preventing Temperature Heat Island A GPU Heterogeneous Cluster Scheduling Model for Preventing Temerature Heat Island Yun-Peng CAO 1,2,a and Hai-Feng WANG 1,2 1 School of Information Science and Engineering, Linyi University, Linyi Shandong,

More information

Geo-activity Recommendations by using Improved Feature Combination

Geo-activity Recommendations by using Improved Feature Combination Geo-activity Recoendations by using Iproved Feature Cobination Masoud Sattari Middle East Technical University Ankara, Turkey e76326@ceng.etu.edu.tr Murat Manguoglu Middle East Technical University Ankara,

More information

Near Light Correction for Image Relighting and 3D Shape Recovery

Near Light Correction for Image Relighting and 3D Shape Recovery Near Light Correction for Iage Relighting and 3D Shae Recovery Anonyous for Review Abstract In this aer, we roose a near-light illuination odel for iage relighting and 3D shae recovery Classic ethods such

More information

Data Caching for Enhancing Anonymity

Data Caching for Enhancing Anonymity Data Caching for Enhancing Anonyity Rajiv Bagai and Bin Tang Departent of Electrical Engineering and Coputer Science Wichita State University Wichita, Kansas 67260 0083, USA Eail: {rajiv.bagai, bin.tang}@wichita.edu

More information

Joint Measurement- and Traffic Descriptor-based Admission Control at Real-Time Traffic Aggregation Points

Joint Measurement- and Traffic Descriptor-based Admission Control at Real-Time Traffic Aggregation Points Joint Measureent- and Traffic Descriptor-based Adission Control at Real-Tie Traffic Aggregation Points Stylianos Georgoulas, Panos Triintzios and George Pavlou Centre for Counication Systes Research, University

More information

Clustering. Cluster Analysis of Microarray Data. Microarray Data for Clustering. Data for Clustering

Clustering. Cluster Analysis of Microarray Data. Microarray Data for Clustering. Data for Clustering Clustering Cluster Analysis of Microarray Data 4/3/009 Copyright 009 Dan Nettleton Group obects that are siilar to one another together in a cluster. Separate obects that are dissiilar fro each other into

More information

Research on a Kind of QoS-Sensitive Semantic Web Services Composition Method Based on Genetic Algorithm

Research on a Kind of QoS-Sensitive Semantic Web Services Composition Method Based on Genetic Algorithm roceedings of the 7th International Conference on Innovation & Manageent 893 Research on a Kind of QoS-Sensitive Seantic Web Services Coposition Method Based on Genetic Algorith Cao Hongjiang, Nie Guihua,

More information

Integrating fast mobility in the OLSR routing protocol

Integrating fast mobility in the OLSR routing protocol Integrating fast obility in the OLSR routing protocol Mounir BENZAID 1,2, Pascale MINET 1 and Khaldoun AL AGHA 1,2 1 INRIA, Doaine de Voluceau - B.P.105, 78153 Le Chesnay Cedex, FRANCE ounir.benzaid, pascale.inet@inria.fr

More information

Multi Packet Reception and Network Coding

Multi Packet Reception and Network Coding The 2010 Military Counications Conference - Unclassified Progra - etworking Protocols and Perforance Track Multi Packet Reception and etwork Coding Aran Rezaee Research Laboratory of Electronics Massachusetts

More information

TensorFlow and Keras-based Convolutional Neural Network in CAT Image Recognition Ang LI 1,*, Yi-xiang LI 2 and Xue-hui LI 3

TensorFlow and Keras-based Convolutional Neural Network in CAT Image Recognition Ang LI 1,*, Yi-xiang LI 2 and Xue-hui LI 3 2017 2nd International Conference on Coputational Modeling, Siulation and Applied Matheatics (CMSAM 2017) ISBN: 978-1-60595-499-8 TensorFlow and Keras-based Convolutional Neural Network in CAT Iage Recognition

More information

IN many interactive multimedia streaming applications, such

IN many interactive multimedia streaming applications, such 840 IEEE TRANSACTIONS ON MULTIMEDIA, VOL. 8, NO. 5, MAY 06 A Geoetric Approach to Server Selection for Interactive Video Streaing Yaochen Hu, Student Meber, IEEE, DiNiu, Meber, IEEE, and Zongpeng Li, Senior

More information

Evaluation of the Timing Properties of Two Control Networks: CAN and PROFIBUS

Evaluation of the Timing Properties of Two Control Networks: CAN and PROFIBUS Evaluation of the Tiing Properties of Two Control Networs: CAN and PROFIBUS Max Mauro Dias Santos 1, Marcelo Ricardo Steer 2 and Francisco Vasques 3 1 UnilesteMG, CEP 35170-056, Coronel Fabriciano MG Brasil.

More information

Single Degree-of-Freedom Rigidly Foldable Cut Origami Flashers

Single Degree-of-Freedom Rigidly Foldable Cut Origami Flashers Robert J. Lang Lang Origai, Alao, CA 9507 e-ail: robert@langorigai.co Sencer Magleby Deartent of Mechanical Engineering, Brigha Young University, Provo, UT 860 e-ail: agleby@byu.edu Larry Howell Deartent

More information

ELEVATION SURFACE INTERPOLATION OF POINT DATA USING DIFFERENT TECHNIQUES A GIS APPROACH

ELEVATION SURFACE INTERPOLATION OF POINT DATA USING DIFFERENT TECHNIQUES A GIS APPROACH ELEVATION SURFACE INTERPOLATION OF POINT DATA USING DIFFERENT TECHNIQUES A GIS APPROACH Kulapraote Prathuchai Geoinforatics Center, Asian Institute of Technology, 58 Moo9, Klong Luang, Pathuthani, Thailand.

More information

Real Time Displacement Measurement of an image in a 2D Plane

Real Time Displacement Measurement of an image in a 2D Plane International Journal of Scientific Research Engineering & Technology (IJSRET), ISSN 2278 0882 Volue 5, Issue 3, March 2016 176 Real Tie Displaceent Measureent of an iage in a 2D Plane Abstract Prashant

More information

A Comparative Study of Two-phase Heuristic Approaches to General Job Shop Scheduling Problem

A Comparative Study of Two-phase Heuristic Approaches to General Job Shop Scheduling Problem IE Vol. 7, No. 2, pp. 84-92, Septeber 2008. A Coparative Study of Two-phase Heuristic Approaches to General Job Shop Scheduling Proble Ji Ung Sun School of Industrial and Managent Engineering Hankuk University

More information

Survivability Function A Measure of Disaster-Based Routing Performance

Survivability Function A Measure of Disaster-Based Routing Performance Survivability Function A Measure of Disaster-Based Routing Perforance Journal Club Presentation on W. Molisz. Survivability function-a easure of disaster-based routing perforance. IEEE Journal on Selected

More information

A Measurement-Based Model for Parallel Real-Time Tasks

A Measurement-Based Model for Parallel Real-Time Tasks A Measureent-Based Model for Parallel Real-Tie Tasks Kunal Agrawal 1 Washington University in St. Louis St. Louis, MO, USA kunal@wustl.edu https://orcid.org/0000-0001-5882-6647 Sanjoy Baruah 2 Washington

More information

cmd p p p cmd cmd ... cmd p p p command 1 command 2 command 3 cmd cmd ... cmd S M T W T F S command n cmd ... cmdn p p p cmd ...

cmd p p p cmd cmd ... cmd p p p command 1 command 2 command 3 cmd cmd ... cmd S M T W T F S command n cmd ... cmdn p p p cmd ... The Trials and Tribulations of Building an Adative User Interface Benjain Korveaker & Russell Greiner fbenjain, greinerg@cs.ualberta.ca Deartent of Couter Science University of Alberta Edonton, Canada

More information

Designing High Performance Web-Based Computing Services to Promote Telemedicine Database Management System

Designing High Performance Web-Based Computing Services to Promote Telemedicine Database Management System Designing High Perforance Web-Based Coputing Services to Proote Teleedicine Database Manageent Syste Isail Hababeh 1, Issa Khalil 2, and Abdallah Khreishah 3 1: Coputer Engineering & Inforation Technology,

More information

A Model-Adaptable MOSFET Parameter Extraction System

A Model-Adaptable MOSFET Parameter Extraction System A Model-Adatable MOSFET Parameter Extraction System Masaki Kondo Hidetoshi Onodera Keikichi Tamaru Deartment of Electronics Faculty of Engineering, Kyoto University Kyoto 66-1, JAPAN Tel: +81-7-73-313

More information

DYNAMIC ESTIMATION OF BDP IN MANETS FOR EFFECTIVE NEXT NODE SELECTION

DYNAMIC ESTIMATION OF BDP IN MANETS FOR EFFECTIVE NEXT NODE SELECTION www.arpnjournals.co DYNAMIC ESTIMATION OF BDP IN MANETS FOR EFFECTIVE NEXT NODE SELECTION N. Snehalatha 1 and Paul Rodrigues 2 1 School of Coputing, SRM University, Chennai, Tail Nadu, India 2 Departent

More information

AGV PATH PLANNING BASED ON SMOOTHING A* ALGORITHM

AGV PATH PLANNING BASED ON SMOOTHING A* ALGORITHM International Journal of Software Engineering & Applications (IJSEA), Vol.6, No.5, Septeber 205 AGV PATH PLANNING BASED ON SMOOTHING A* ALGORITHM Xie Yang and Cheng Wushan College of Mechanical Engineering,

More information

Oblivious Routing for Fat-Tree Based System Area Networks with Uncertain Traffic Demands

Oblivious Routing for Fat-Tree Based System Area Networks with Uncertain Traffic Demands Oblivious Routing for Fat-Tree Based Syste Area Networks with Uncertain Traffic Deands Xin Yuan Wickus Nienaber Zhenhai Duan Departent of Coputer Science Florida State University Tallahassee, FL 3306 {xyuan,nienaber,duan}@cs.fsu.edu

More information

Enhancing Real-Time CAN Communications by the Prioritization of Urgent Messages at the Outgoing Queue

Enhancing Real-Time CAN Communications by the Prioritization of Urgent Messages at the Outgoing Queue Enhancing Real-Tie CAN Counications by the Prioritization of Urgent Messages at the Outgoing Queue ANTÓNIO J. PIRES (1), JOÃO P. SOUSA (), FRANCISCO VASQUES (3) 1,,3 Faculdade de Engenharia da Universidade

More information

Handling Indivisibilities. Notes for AGEC 622. Bruce McCarl Regents Professor of Agricultural Economics Texas A&M University

Handling Indivisibilities. Notes for AGEC 622. Bruce McCarl Regents Professor of Agricultural Economics Texas A&M University Notes for AGEC 6 Bruce McCarl Regents Professor of Agricultural Econoics Texas A&M University All or None -- Integer Prograing McCarl and Spreen Chapter 5 Many investent and probles involve cases where

More information

Verifying the structure and behavior in UML/OCL models using satisfiability solvers

Verifying the structure and behavior in UML/OCL models using satisfiability solvers IET Cyber-Physical Systes: Theory & Applications Review Article Verifying the structure and behavior in UML/OCL odels using satisfiability solvers ISSN 2398-3396 Received on 20th October 2016 Revised on

More information

An Efficient Coding Method for Coding Region-of-Interest Locations in AVS2

An Efficient Coding Method for Coding Region-of-Interest Locations in AVS2 An Efficient Coding Method for Coding Region-of-Interest Locations in AVS2 Mingliang Chen 1, Weiyao Lin 1*, Xiaozhen Zheng 2 1 Deartment of Electronic Engineering, Shanghai Jiao Tong University, China

More information

Abstract. for L1VPNs [2, 3] and the standardization work at the Internet Engineering Task Force (IETF) is in progress [4].

Abstract. for L1VPNs [2, 3] and the standardization work at the Internet Engineering Task Force (IETF) is in progress [4]. Traffic Deand Modeling for Dynaic Layer 1 VPN Serices * Joachi Scharf, Martin Köhn Uniersity of Stuttgart, Institute of Counication Networks and Coputer Engineering (IKR), Pfaffenwaldring 47, 70569 Stuttgart,

More information

Short Papers. Crosstalk in VLSI Interconnections

Short Papers. Crosstalk in VLSI Interconnections IEEE TRANSACTIONS ON COPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTES, VOL. 8, NO. 2, DECEBER 999 87 Short Papers Crosstalk in VLSI Interconnections Ashok Vittal, Lauren Hui Chen, algorzata arek-sadowska,

More information