Dynamic Multiple Parity (DMP) Disk Array for Serial Transaction Processing

Size: px
Start display at page:

Download "Dynamic Multiple Parity (DMP) Disk Array for Serial Transaction Processing"

Transcription

1 IEEE TRANSACTIONS ON COMPUTERS, VOL. 50, NO. 9, SEPTEMBER Dynamic Multiple Paity (DMP) Disk Aay fo Seial Tansaction Pocessing K.H. Yeung, Membe, IEEE, and T.S. Yum, Senio Membe, IEEE AbstactÐThe pefomance of today's database systems is usually limited by the speed of thei I/O devices. Fast I/O systems can be built fom an aay of low cost disks woking in paallel. This kind of disk achitectue is called RAID (Redundant Aays of Inexpensive Disks). RAID pomises impovement ove SLED (Single Lage Expensive Disks) in pefomance, eliability, powe consumption, and scalability. Howeve, a geneal fact about RAID is that the ªwiteº opeation is difficult to speedup. In this pape, we popose a new RAID achitectue, called Dynamic Multiple Paity (DMP) Disk Aay, fo seial tansaction pocessing database systems. Seial tansaction pocessing database systems include engineeing database systems, fully eplicated database systems using a completely centalized algoithm and distibuted systems using the consevative timestamp odeing algoithm. DMP Disk Aay can significantly incease the I/O thoughput by incopoating multiple paity disks. Due to the inheent distibuted spaing popety, DMP Disk Aay can povide nomal sevice to the uses unde single disk failue condition. Delay and maximum thoughput analysis on DMP Disk Aay is pefomed. Results show that, fo a typical ªwiteº job popotion of 20 pecent, DMP Disk Aay can povide nealy 20 pecent impovement on I/O thoughput ove that of RAID level 5 when one exta paity disk is used. Index TemsÐRAID, disk aays, I/O systems, database systems, tansaction pocessing. æ INTRODUCTION IN the past decade, consideable attention has been dawn to the eseach and development of database computes. As Su [] has stated in his book, thee ae thee easons why database computes ae needed: ) the need fo efficient and effective data management, 2) the need fo moe poweful database management systems, and 3) high pefomance database computes become economically feasible because of the advancement in hadwae technology and the eduction in hadwae cost. The pefomance of a database system is usually limited by the speed of its stoage devices. One good example is the Datacycle2 poject at BellCoe stating in the late 980s [2]. BellCoe poposed an innovative database achitectue called the Datacycle2. In this achitectue, the entie database is peiodically pumped out fom the cental database to a numbe of seves in which the use equied data ae filteed out. Since the whole database is pumped out, Datacycle2 has an unlimited thoughput fo ead-only tansactions. Datacycle2 technique assumes that the database is ªmemoy esident,º i.e., the entie database has to eside in vey fast stoage. This assumption, howeve, limits its scope of applications. Redundant Aays of Inexpensive Disks (RAID) is an innovative concept in designing fast and eliable data stoage systems. The philosophy behind RAID is that,. In the pototype built by BellCoe, the whole database esides in RAM.. K.H. Yeung is with the Depatment of Electonic Engineeing, City Univesity of Hong Kong, Yau Yat Chuen, Hong Kong. eeayeung@cityu.edu.hk.. T.S. Yum is with the Depatment of Infomation Engineeing, Chinese Univesity of Hong Kong, Shatin, Hong Kong. yum@ie.cuhk.edu.hk. Manuscipt eceived 2 Aug. 994; accepted 26 Feb Fo infomation on obtaining epints of this aticle, please send to: tc@compute.og, and efeence IEEECS Log Numbe instead of using one single expensive disk to achieve the pefomance and eliability equied, an aay of low cost disks woking in paallel ae used. Five levels of RAID wee defined when RAID was fist intoduced 2 and RAID level 5 was found to be one of the best [3]. Fo all levels of RAID, the ªwiteº opeations ae much slowe than the ªeadº ones. This limitation is paticulaly sevee fo applications with fequent data updates. Thee ae two easons why a ªwiteº opeation takes moe time fo RAID. Fist, a ªwiteº opeation involves the additional step of eading back the old data fom one disk and paity fom anothe disk. Second, a ªwiteº opeation involves the waiting time fo two specific disks to be fee simultaneously befoe actual witing. This waiting time can be educed by using the technique pesented in this pape. In this pape, we popose a new RAID achitectue, called Dynamic Multiple Paity (DMP) Disk Aay, fo seial tansaction pocessing database systems. Many database systems pocess tansactions in this way. Examples ae engineeing database systems [4], [5], [6], fully eplicated database systems using the completely centalized algoithm, and distibuted systems using the consevative timestamp odeing algoithm [7], [8], [9]. When we discuss the opeation of DMP Disk Aays, we will see how this kind of database systems handles I/O equests in a way diffeent fom othe database systems. We will also show that DMP Disk Aay can significantly educe the waiting time of disk opeations and povide a highe I/O thoughput than RAID level 5. In the next section, we descibe DMP Disk Aay in detail. We then pesent an aveage delay analysis in Section 3 and a maximum thoughput analysis in Section 4. In Section 5, esults of simulation with a pecise disk model ae given. We then conclude the pape in Section Some disk manufactos intoduced thei own levels of RAID late /0/$0.00 ß 200 IEEE

2 950 IEEE TRANSACTIONS ON COMPUTERS, VOL. 50, NO. 9, SEPTEMBER 200 Fig.. The secto coodinate system showing level 5 RAID. Fig. 2. The placement of paity sectos fo DMP Disk Aay when R ˆ 2. 2 DMP DISK ARRAY 2. Secto Coodinate System The secto coodinate system can be used to descibe RAID opeations, including that of DMP Disk Aay. Conside a disk aay system with M disks, whee each disk consists of Z sectos and each secto can stoe K bits of infomation (Fig. ). Let S i; j ˆ b i;j ;bi;j 2 ;bi;j 3 ;...;bi;j K be the bit patten of secto j in disk i. It can be data o paity. As an example, a RAID level is descibed in this secto coodinate system as: S i; j ˆS i ;j j ˆ ; 2; 3;...;Z; iˆ ; 3; 5;...;M : Similaly, RAID level 4 and level 5 can be descibed as: X M iˆ S i; j ˆ ; ; ;...; fo all j fodd paityg 0; 0; 0;...; 0 fo all j feven paityg; whee is defined hee as the mod-2 sum of the sectos' contents: X M iˆ S ; j ˆ S i; j S 2; j... S M; j 2 ˆ b ;j ;b;j 2 ;...;b;j K b2;j ;b2;j 2 ;...;b2;j K... b M;j ;b M;j 2 ;...;b M;j K ˆ b ;j b 2;j ;b ;j 2 b 2;j 2 2 ;...; b ;j K b2;j K... bm;j K : 3 Let E j be the location of the paity secto at ow j. Fo RAID level 4 E j ˆM and fo RAID level 5 E j ˆ j mod MŠ. The placement of these paity sectos is shown in Fig.. A study of the vaious paity placement methods fo RAID level 5 can be found in [0]. 2.2 Secto Oganization The DMP Disk Aay poposed in this pape is a new RAID achitectue fo which RAID level 5 is a special case. RAID level 5 has only one paity secto in each secto ow and, theefoe, it is not possible to simultaneously update two o moe sectos on the same ow. DMP Disk Aay allows such updates by placing R paity sectos in each ow. Thei locations E j ;E 2 j ;...;E R j fo ow j ae E j ˆ j 2 mod MŠ ˆ ; 2;...;R: 4 Lee and Katz [0] showed that, fo elatively lage equest sizes of hundeds of kilobytes, the choice of paity placement can significantly affect the pefomance of disk aays, wheeas, fo small equest sizes, the choice of paity placement is insignificant to system pefomance. We theefoe abitaily choose the paity locations as stated in (4). Although thee ae R paity sectos in each ow, paity integity descibed in (2) is always maintained fo DMP Disk Aay. (An example on the secto oganization of DMP Disk Aay fo R ˆ 2 is shown in Fig. 2.) Thee ae two advantages to placing R paity sectos in each ow. Fist, fo each data secto modification, we can choose any one of the R paity sectos in the same ow fo simultaneous paity modification. Hence, blocking due to busy disks can be significantly educed. Second, up to R data sectos on the same ow can now be modified simultaneously. We now pove that DMP Disk Aay has these useful popeties. 2.3 Popeties of DMP Disk Aay Popety concens the simultaneous negation of two bits on the same ow. Popety. Fo any ow j, simultaneous negation of any two bits in the same bit positions of two diffeent sectos will not affect the paity sum of the ow. Poof. Conside the set of th bits of each secto in secto ow j, i.e., b ;j, b2;j,..., and bm;j. In ode to maintain paity integity, the sum of these bits should always be fo odd paity o 0 fo even paity. The paity sum afte the negation of any two bits b a;j and b b;j, whee a 6ˆ b is:

3 YEUNG AND YUM: DYNAMIC MULTIPLE PARITY (DMP) DISK ARRAY FOR SERIAL TRANSACTION PROCESSING 95 b ;j... b a;j... b b;j b b;j b ;j fcommutative lawg b b;j b ;j fassociative lawg b b;j b a;j b b;j b ;j fdefinition of exclusive-og bb;j b a;j b b;j b ;j flaw of double negationg b b;j b ;j fdefinition of exclusive-og b b;j b ;j fassociative lawg ˆ b ;j... b a;j fcommutative lawg:... b b;j 5 tu secto update equies the old contents of two sectos only. Theefoe, each update is actually caied out by two disks woking in coopeation and is independent of the opeations of the est of the disks. Theefoe, with R paity sectos, R simultaneous updates can be pefomed. tu Popety 5. Conside a DMP Disk Aay with M disks and R paity sectos in each ow (R >). When a disk fails, it is possible to econfigue the emaining M disks to a new aay with R paity sectos in each ow without data loss. Poof. Let (p, p,..., p) be the paity sum of all the sectos in a ow, say ow j, and let S b ;j ;S b 2 ;j ;...;S b R ;j be the paity sectos. Obviously, b ˆ E j ;b 2 ˆ E 2 j ;...;b R ˆ E R j : Suppose disk i fails. We conside two cases fo the ecovey of S i; j.. If S i; j is a data secto, it can be ecoveed fom S i; j ˆ p; p;...;p XM nˆ;n6ˆi S n; j : 8 The last esult is identical to the paity sum befoe the simultaneous negation. We can extend the agument on the modification of two bits to that of the modification of two sectos on the same ow. This is stated as Popety 2. Popety 2. Paity integity of a ow can be maintained by modifying any one of the paity sectos in the same ow. Popety 2 implies that thee can be R diffeent ways to update a data secto. Due to this flexibility, the pobability that a ªwiteº equest is blocked due to the busy disk can be educed. Popety 3. Conside the simultaneous modification of data in secto j of disk a and paity in secto j of disk b. We denote the old data secto, the new data secto, the old paity secto, and the new paity secto as S a; j j old, S a; j j new, S b; j j old, and S b; j j new, espectively. Fo maintaining data integity, the new paity secto should be: S b; j j new ˆ S a; j j old S a; j j new S b; j j old : Poof. In ode to maintain paity integity, the patial sum of the two sectos S a; j and S b; j must not be changed afte secto modification, i.e., S a; j j new S b; j j new ˆ S a; j j old S b; j j old : Solving fo S b; j j new, (6) is obtained. Popety 4. Fo DMP Disk Aay with R paity sectos in each ow, R data sectos located in the same ow can be simultaneously updated. Poof. Equation (7) shows that, fo any secto update on ow j, the patial sum of the data secto and the paity secto will always be the same. Thus, the paticula secto update will not affect the updating of the othe sectos in ow j. Fom Popety 3, we find that each 6 7 tu One of the paity sectos, say S b R ;j, can be used to stoe the ecoveed data S i; j. To maintain paity integity anothe paity secto, say S b ;j, is modified as: S b ;j ˆS b ;j S b R ;j : 9 The ecoveed DMP Disk Aay now has R paity sectos. 2. If S i; j is a paity secto, then no data is lost. The paity integity can be ecoveed by modifying anothe paity secto, say b, as follows: S b; j ˆ p; p;...;p XM nˆ;n6ˆi; n6ˆb S n; j : 0 Popety 5 tell us that DMP Disk Aay has the distibuted spaing popety discussed in []. It is shown in [] that distibuted spaing is the best spaing technique fo small disk aays. 2.4 Pinciple of Opeation Fig. 3 shows the oganization of DMP Disk Aay. Requests fom a host ae diectly sent to the disk contolle fo the I/O opeations. The disk contolle consists of fou pats: an FCFS queue, a local memoy, a scheduling pocesso, and a DMA contolle. The queue is used fo stoing I/O equests. Since we ae consideing systems which execute tansactions in a stict ode, a single global queue with FCFS sevice discipline is used. Note that this is diffeent fom othe systems epoted in the liteatue, which use sepaate disk queues [2], [3], [4], [5]. The data associated with each equest (i.e., the new data fo a secto) is stoed in the local memoy when the equest is placed on the queue. The local memoy is also used fo buffeing the data sent to/ead fom each of the disks tu

4 952 IEEE TRANSACTIONS ON COMPUTERS, VOL. 50, NO. 9, SEPTEMBER 200 Fig. 3. Oganization of DMP Disk Aay. Fig. 4. Queuing models of DMP Disk Aay. with the help of the DMA contolle. The DMA contolle functions basically as a multiplexe/demultiplexe and handles simultaneous data tansfes to vaious disks. The scheduling pocesso is esponsible fo distibuting I/O equests to the disks and pefoms all necessay pocessing. Specifically, its functions ae outlined as follows:. It eads a equest fom the FCFS queue when eady and detemines whethe the equest is a ªeadº o a ªwiteº type. 2. Fo a ªeadº equest, the pocesso will a. check the status of the disk involved with this equest; b. instuct the disk involved to ead the taget secto; c. load the secto to the local memoy; and d. signal the host fo data eady. 3. Fo a ªwiteº equest, the pocesso will a. check the status of the disks and select at andom the paity secto of a nonbusy disk; b. ead the old data secto and the selected paity secto; c. compute the new paity secto accoding to (7); d. wite the new data secto and the new paity secto to thei coesponding disks; e. ead back the paity and data sectos fo veification; and f. signal the host fo ªwiteº completion. 3 AVERAGE DELAY 3. Analysis Fig. 4a shows a queuing model fo DMP Disk Aay. Requests sent fom host become jobs to be seved in the seves. Job aivals ae assumed to be a Poisson pocess with ate. A job is of the ªwiteº type with pobability and of the ªeadº type with the emaining pobability. Jobs not yet seved by the disk aay ae queued in an FCFS

5 YEUNG AND YUM: DYNAMIC MULTIPLE PARITY (DMP) DISK ARRAY FOR SERIAL TRANSACTION PROCESSING 953 queue. We call the job which is at the top of the queue the Head Of Line (HOL) job. The pobability that the HOL job needs to access a paticula disk is assumed to be the same fo all disks. This assumption is usually not tue, but it can be made tue by distibuting the fequently accessed data unifomly acoss all the disks. Fo mathematical convenience, we assume that the sevice time fo a job is exponentially distibuted with the sevice ates fo a ªwiteº job and a ªeadº job denoting as w and, espectively. Let andom vaiables N w and N denote the numbe of ªwiteº jobs and ªeadº jobs in the disk aay and N q be the numbe of queuing jobs (including the HOL job) at any time. It is easy to see that the tiplet N w ;N ;N q completely specifies the state of the system. Let S w;;q denote the state of the system when N w ˆ w, N ˆ, and N q ˆ q. State tansition will take place when a new job aives o when a job in the system depats. Since the time spent in a state is exponentially distibuted, the evolution of N w ;N ;N q is a continuous time Makov pocess. We define the tansition pobabilities to be P S w 0 ; 0 ;q 0 j S w;;q ˆ P N w ;N ;N q ˆ w 0 ; 0 ;q 0 afte state tansition j N w ;N ;N q ˆ w; ; q befoe state tansition : Conside a paticula state tansition at time t. Define events E, E 2, and E 3 as: E : A new job aives at time t. E 2 : A ªeadº job depats at time t. E 3 : A ªwiteº job depats at time t. The event E is listed in the fist column of Table. A complete table listing the othe two events is given in [3]. The disk aay is at S w;;q immediately befoe t, i.e., at time t t (t! 0). The pobability that a new job will aive in the inteval t t; t is t if t! 0. Similaly, the pobabilities that the disk aay will finish seving a ªwiteº job and a ªeadº job in this small time inteval ae w t and t, espectively. Theefoe, P E Šˆ w w P E 2 Šˆ 2 w w w w P E 3 Šˆ : w w When a new job aives (i.e., E occus), the pobabilities that this HOL job is of the ªeadº type and of the ªwiteº type ae and, espectively. Howeve, a diffeent pobability fo each of the job types fo the HOL job is found when a job depats (i.e., eithe E 2 o E 3 occus). We denote the pobabilities that the HOL job is of the ªwiteº type and of the ªeadº type by h w and h w, espectively, fo system tansitions due to job depatues. Pobability h w can be deived as follows: At the pevious state change, the HOL job was blocked because it equies the access of one o moe busy disks. By that time thee wee 2w busy disks. If the HOL job is of the ªeadº type, the pobability of blocking k is: numbe of busy disks k ˆ total numbe of disks ˆ 2w M : 3 On the othe hand, if the HOL job is of the ªwiteº type, the pobability of blocking k w can be deived as follows: We shall call the disk which the HOL job tagets fo data modification the data disk and the R disks stoing the equied paity infomation the paity disks. Let events and 2 be: : The data disk to be accessed was fee at the pevious state change. 2 : At least one of the R paity disks was fee at the pevious state change. Then, k w ˆ P no blockingš ˆ P and 2 Š ˆ P ŠP 2 j Š: P 2 j Š is given by: 0 numbe of ways to B choose R paity disks A fom 2w busy disks P 2 j Šˆ 0 numbe of ways to B choose R paity disks A fom M disks 8 2w <R >< 2w ˆ R 2w R: M >: R 4 5 Substituting into (4) and solving fo k w we obtain 8 2w 2w <R M >< 2w k w ˆ 2w M 2w R 2w R: M M M >: R 6 Having obtained k and k w, h w is given by: h w ˆ kw k k w : 7 Conditioning on event E i and giving the type of the HOL job, the pobabilities of having diffeent numbes of jobs located in the queue and in the disk aay ae listed in the sixth column of Table. Fo example, the fist ow of Table coesponds to the case that a ªeadº job entes an empty queue is blocked o the numbe of ªeadº jobs and ªwiteº jobs in the disk aay emain the same and q 0 becomes. The pobability of having the new tiplet w 0 ; 0 ;q 0 ˆ w; ; afte t unde the two given conditions is denoted as a. Similaly the fifth ow of Table coesponds to the case that a ªwiteº job entes an empty queue and gets seved immediately. The pobability of having w 0 ; 0 ;q 0 ˆ w ;;0 afte t unde

6 954 IEEE TRANSACTIONS ON COMPUTERS, VOL. 50, NO. 9, SEPTEMBER 200 TABLE Tansition Pobabilities of Event E fo Aveage Delay Analysis A complete table showing the tansition pobabilities of all othe events can be found in [3]. the two given conditions is denoted by a 3, as shown. The deivation of all a i s is given in [3]. The seventh column of Table shows the type of the new HOL job and its coesponding pobability. A new HOL job is blocked at time t with pobabilities k 0 and k 0 w when it is of the ªeadº type and of the ªwiteº type, espectively. Similaly to the deivations on k and k w, k 0 and k 0 w can be obtained fom (3) and (6) by changing the numbe of busy disks fom 2w to 2w 0 0. The last column of Table shows P S w 0 ; 0 ;q 0jS w;;q;e i Š.By emoving the condition on E i, the tansition pobabilities ae thus obtained: P S w 0 ; 0 ;q 0 j S X 3 w;;q ˆ iˆ P S w 0 ; 0 ;q 0 j S w;;q;e i P Ei Š: 8 Having obtained the tansition pobabilities, the equilibium distibution of diffeent states can be computed in the usual way. The expected numbes of jobs in the queue ae given by: E N q Šˆ X bm=2c X M X wˆ0 ˆ0 qˆ0 qp S w;;q Š: 9 Finally, by Little's fomula, the job's sojoun time D is given by D ˆ E N qš : 20 w 3.2 Numeical Example As an example, conside a small DMP Disk Aay with data stoage capacity of fou disks. We assume in this example that w ˆ 30 jobs/sec and ˆ 50 jobs/sec. Figs. 5 and 6 show both the analytic and simulation esults fo this disk aay and we obseve that they match vey well with each othe. Note that, fo all simulation esults shown in this pape, we have extended the simulation time sufficiently long to make the 95 pecent confidence intevals smalle than the size of the simulation points shown. Fig. 5 shows the job delay against the aival ate when half of the jobs ae of the ªwiteº type. As indicated by the cuve, the maximum thoughput fo RAID level 5 is about 50 jobs/sec. When one paity disk is added to the disk aay (i.e., when M ˆ 6 and R ˆ 2), we find that the aveage job delay is educed unde all taffic conditions and the maximum thoughput is inceased by about 3 pecent when compaed to RAID level 5. If one moe paity disk is used (R ˆ 3), we find that the job delay is futhe educed unde all taffic conditions and 23 pecent incease in maximum thoughput is obseved. Fig. 6 shows the job delay against the aival ate when 20 pecent of the jobs ae of the ªwiteº type ( ˆ 0:2). We obseve fom the figue that DMP Disk Aay with R ˆ 2 and R ˆ 3 again pefoms significantly bette than RAID level 5 unde all taffic levels. 4 MAXIMUM THROUGHPUT 4. Analysis Although the analysis given in Section 3 povides an exact solution on aveage job delay, the computation is vey demanding when the disk aay is lage. In the following, we pesent a simplified analysis which gives the maximum thoughput fo DMP Disk Aay. A modified model, shown in Fig. 4b, is used in ou analysis. Compaed with the pevious model (Fig. 4a), the FCFS queue is emoved and

7 YEUNG AND YUM: DYNAMIC MULTIPLE PARITY (DMP) DISK ARRAY FOR SERIAL TRANSACTION PROCESSING 955 Fig. 5. Aveage job delay against the aival ate fo small DMP Disk Aays (data stoage capacity M-R is fou disks). Half the jobs ae of ªwiteº type ( ˆ 0:5). Fig. 6. Aveage job delay against the aival ate fo small DMP Disk Aays (data stoage capacity M-R is fou disks). Eighty pecent of the jobs ae of the ªeadº type ( ˆ 0:2). we assume that thee is always a new job available at the input of the disk aay. All othe peviously used assumptions ae used in this maximum thoughput analysis. Since the queue is emoved, the system's state can solely be specified by N w and N and is denoted by S w;. State tansition will take place when a job in the disk aay depats, i.e., eithe E 2 o E 3 occus. Since the time spent in a state is exponentially distibuted, the evolution of N w and N emains a continuous time Makov pocess. As befoe, we define the tansition pobabilities to be P S w; ;S w0 ; 0Šˆ P N w ˆ w 0 ;N ˆ 0 afte system tansition j N w ˆ w; N ˆ befoe system tansitionš: 2 Conside a paticula state tansition occus at time t. The pobability of occuence of E 2 and E 3 is given by P E 2 Šˆ w w w w P E 3 Šˆ : w w 22 Given that a specific event E i occus, the tansition pobabilities P S w; ;S w 0 ; 0jE iš ae listed in the last column of Table 2. 3 The deivations on pobabilities a i shown in the table ae given in [3]. Having obtained the tansition pobabilities, the equilibium distibution of diffeent states can be computed as befoe. The expected time between successive job depatues X is given by 3. Table 2 shows only the tansition pobabilities of event E. A complete table fo all events can be found in [6].

8 956 IEEE TRANSACTIONS ON COMPUTERS, VOL. 50, NO. 9, SEPTEMBER 200 TABLE 2 Tansition Pobabilities of Event E 2 fo Maximum Thoughput Analysis A complete table showing the tansition pobabilities of all othe events can be found in [3]. Fig. 7. Thoughput gain ove that of RAID level 5 as a function of R fo small DMP Disk Aays (data stoage capacity M-R is fou disks). bm=2c X ˆ X wˆ0 X M ˆ0 P S w; Š : w w 23 Finally, the maximum thoughput fo DMP Disk Aays T is T ˆ X : Numeical Examples Fig. 7 shows the thoughput gain ove that of RAID level 5 against R fo a small DMP Disk Aay with data stoage capacity of fou disks. We obseve fom the figue that inceasing the numbe of paity disks R will always incease the maximum thoughput fo DMP Disk Aay. When the popotion of ªwiteº jobs is highe, the incease in

9 YEUNG AND YUM: DYNAMIC MULTIPLE PARITY (DMP) DISK ARRAY FOR SERIAL TRANSACTION PROCESSING 957 Fig. 8. Thoughput gain ove that of RAID level 5 as a function of R fo lage DMP Disk Aays (data stoage capacity M-R is 3 disks). maximum thoughput is moe appaent because DMP Disk Aay will educe the queuing time of ªwiteº jobs. Consideing the case of all ªwiteº jobs ( ˆ ), DMP Disk Aay with R ˆ 2 povides 24 pecent incease in maximum thoughput when compaed to RAID level 5. Futhe inceasing R to thee povides an additional 7 pecent incease in maximum thoughput. These pefomance figues match well with the numeical examples given in the pevious section. When R is geate than thee, linea incease in maximum thoughput is obseved fo each paity disk added. Fom the figue, we also obseve that significant thoughput gain is obtained fo a typical 20 pecent popotion of ªwiteº jobs. Fig. 8 plots the thoughput gain ove that of RAID level 5 against R fo a lage DMP Disk Aay with data stoage capacity of 3 disks. When compaed with Fig. 7, we obseve that DMP Disk Aay with R ˆ 2 povides even moe notable incease in maximum thoughput than the pevious case. We can thus conclude that a DMP Disk Aay with R ˆ 2 povides the best cost/pefomance atio. 5 SIMULATION WITH PRECISE DISK MODEL In ou pevious analysis, disk sevice time is assumed to be exponentially distibuted. This is usually not tue fo pactical disk dives. To bette undestand the pefomance of DMP Disk Aay in pactice, we pefom simulation on DMP Disk Aay with a pecise disk model. In ou simulation, disks ae not assumed to be otationally synchonized and thei simulation paametes ae summaized in Table 3. Each disk access involves a seek time, a latency, and a data tansfe time [6]. We use the seek pofile in [7], which states that the seek time Tseek (in msec) is elated to seek distance x (in numbe of cylindes) by: T seek ˆ 0 p fo x ˆ 0 0:4623 x 0:0092 x 2 fo x>0: 25 Latency is assumed to be unifomly distibuted. Data tansfe time fo one secto is equal to the disk evolution time divided by the numbe of sectos pe tack, as given in Table 3. With that, the mean sevice time fo ªwiteº jobs is computed to be 33.3 ms and, fo ªeadº jobs, it is 20 ms. The coesponding sevice ates ae theefoe the same as those in the pevious examples. As stated in [8], this kind of disk modeling povides moe than 94 pecent accuacy when ignoing the disk caching effect. Since disk caching has little impact on ªwiteº pefomance (which we ae most TABLE 3 Disk Paametes Used in Simulation Fig. 9. Simulation esults on aveage job delay against aival ate fo small disk aays (data stoage capacity M-R is fou disks). Half the jobs ae ªwiteº ( ˆ 0:5).

10 958 IEEE TRANSACTIONS ON COMPUTERS, VOL. 50, NO. 9, SEPTEMBER 200 Fig. 0. Simulation esults on aveage job delay against aival ate fo small disk aays (data stoage capacity M-R is fou disks). Eighty pecent of the jobs ae of the ªeadº type ( ˆ 0:2). inteested in), we can thus assume that the system has no disk caching mechanism. Figs. 9, 0,, and 2 show the simulation esults with the pecise disk model. We fist conside a small disk aay with data stoage capacity of fou disks and half the jobs ae of the ªwiteº type. Fig. 9 shows that the maximum thoughput fo RAID level 5 is about 56 jobs/sec. If DMP Disk Aay with M ˆ 6 and R ˆ 2 is used, the aveage job delay is educed unde all taffic conditions and the maximum thoughput is inceased by 23 pecent as compaed to RAID level 5. The maximum thoughput is futhe inceasted by about 40 pecent when DMP Disk Aay with M ˆ 7 and R ˆ 3 is used. We also obseve that when the numbe of disks M is fixed, DMP Disk Aay with R ˆ 2 still povides pecent impovement in maximum thoughput. Fig. 0 shows the esults fo ˆ 0:2. We obseve that DMP Disk Aay with R ˆ 2 and R ˆ 3 povides 7 pecent and 29 pecent inceases in I/O thoughput, espectively. The incease in thoughput when M ˆ 5 and R ˆ 2 is about 7 pecent. Fig. 2. Simulation esults on aveage job delay against aival ate fo lage disk aays (data stoage capacity M-R is 3 disks). Eighty pecent of the jobs ae of the ªeadº type ( ˆ 0:2). Figs. and 2 show the delay thoughput chaacteistics of a typical lage disk aay with data stoage capacity of 3 disks. DMP Disk Aay again povides significant I/O thoughput incease. When half the jobs ae of the ªwiteº type (Fig. ), the incease on maximum thoughput fo DMP Disk Aay with R ˆ 2 is about 2 pecent, wheeas the coesponding incease in system cost is at most 7 pecent. Unde the condition of 20 pecent of the jobs ae of the ªwiteº type (Fig. 2), we find that the 7 pecent incease in system cost can still give 2 pecent highe thoughput. Figs. 3 and 4 show the case when the numbe of disks, M, is fixed at 4 disks. When half the jobs ae of the ªwiteº type (Fig. 3), the incease on maximum thoughput fo DMP Disk Aay with R ˆ 2 is about 7 pecent. Note that thee is no incease in system cost. Unde the condition of 20 pecent of the jobs ae of the ªwiteº type (Fig. 4), we find that DMP Disk Aay with R ˆ 2 still povides 8 pecent highe thoughput at no additional system cost. Fig.. Simulation esults on aveage job delay against aival ate fo lage disk aays (data stoage capacity M-R is 3 disks). Half the jobs ae of ªwiteº type ( ˆ 0:5). Fig. 3. Simulation esults on aveage job delay against aival ate fo lage disk aays (total numbe of disks, M, is 4). Half the jobs ae of ªwiteº type ( ˆ 0:5).

11 YEUNG AND YUM: DYNAMIC MULTIPLE PARITY (DMP) DISK ARRAY FOR SERIAL TRANSACTION PROCESSING 959 [5] A.L. Naasimha Reddy and P. Banejee, ªAn Evaluation of Multiple-Disk I/O Systems,º IEEE Tans. Computes, vol. 38, no. 2, Dec [6] S.W. Ng, ªAdvances in Disk Technology: Pefomance Issues,º Compute, vol. 3, no. 5, pp. 75-8, May 998. [7] E.K. Lee and R.H. Katz, ªThe Pefomance of Paity Placements in Disk Aays,º IEEE Tans. Computes, vol. 42, no. 6, pp , June 993. [8] C. Ruemmle and J. Wilkes, ªAn Intoduction to Disk Dive Modeling,º Compute, vol. 27, no. 3, pp. 7-28, Ma Fig. 4. Simulation esults on aveage job delay against aival ate fo lage disk aays (total numbe of disks, M, is 4). Eighty pecent of the jobs ae of ªeadº type ( ˆ 0:2). 6 CONCLUSIONS In this pape, we popose a new RAID achitectue, called Dynamic Multiple Paity (DMP) Disk Aay, fo fast database system applications. The DMP Disk Aay povides significant impovement on I/O thoughput ove the RAID level 5. The DMP Disk Aay also inheently has the spaing popety so that it has a highe suvivability unde disk failue conditions. Delay and maximum thoughput analysis on DMP Disk Aay is pefomed. Simulation with pecise disk model shows that, fo a typical ªwiteº job popotion of 20 pecent, DMP Disk Aay can povide nealy 20 pecent impovement on I/O pefomance ove that of RAID level 5 when one exta paity disk is used. REFERENCES [] S.Y.W. Su, Database Computes. McGaw-Hill, 988. [2] T. Bowen, G. Gopal, G. Heman, and W. Mansfield, ªA Scale Database Achitectue fo Netwok Sevices,º IEEE Comm. Magazine, vol. 29, no., Jan. 99. [3] K.H. Yeung, ªHigh Pefomance Disk Aay Achitectues,º PhD dissetation, The Chinese Univ. of Hong Kong, 995. [4] J.L. Encanacao and P.C. Lockemann, Engineeing Databases. Spinge-Velag, 990. [5] K.H. Yeung and T.S. Yum, ªDynamic Paity Logging Disk Aays fo Engineeing Database Systems,º IEE Poc.ÐComputes and Digital Techniques, vol. 44, no. 5, pp , Sept [6] R.H. Katz, Infomation Management fo Engineeing Design, Spinge-Velag, 985. [7] M. Singhal and N.G. Shivaati, Advanced Concepts in Opeating Systems, chapte 20. McGaw-Hill, 994. [8] F.R. McFadden, J.A. Hoffe, and M.B. Pescott, Moden Database Management, fifth ed. Addison-Wesley, 999. [9] S. Cei and G. Pelagatti, Distibuted Databases: Pinciples and Systems, chapte 8. McGaw-Hill, 984. [0] E.K. Lee and R.H. Katz, ªThe Pefomance of Paity Placements in Disk Aays,º IEEE Tans. Computes, vol. 42, no. 6, June 993. [] J. Menon and D. Mattson, ªCompaison of Spaing Altenatives fo Disk Aay,º Poc. 9th Int'l Symp. Compute Achitectue, pp , May 992. [2] M.Y. Kim, ªSynchonized Disk Inteleaving,º IEEE Tans. Computes, vol. 35, no., pp , Nov [3] N.C. Wilhelm, ªA Geneal Model fo the Pefomance of Disk Systems,º J. ACM, vol. 24, no., pp. 4-3, Jan [4] S.W. Ng, ªImpoving Disk Pefomance via Latency Reduction,º IEEE Tans. Computes, vol. 40, no., pp , Jan. 99. K.H. Yeung eceived the BSc degee fom the Chinese Univesity of Hong Kong in 984, the postgaduate diploma and MSc degee fom the City Univesity of Hong Kong in 989 and 99, espectively, and the PhD degee fom the Chinese Univesity of Hong Kong in 995. He was an R&D enginee with a PC manufactue fo one yea befoe joining the City Univesity of Hong Kong in 985. He was an analyst pogamme fo five yeas befoe tansfeing to the Depatment of Electonic Engineeing in 990, whee he is cuently an associate pofesso. His eseach inteests include Intenet system design, mobile communication, infomation delivey, and VoD achitectues. Besides eseach, he is also active in poviding consultancy sevices to local industy in the aeas of compute netwoking, communication systems, and Intenet systems. He is also a Cetified Cisco Academy Instucto. He is a membe of the IEEE. T.S. Yum is a gaduate of Columbia Univesity. He woked at Bell Telephone Laboatoies fo two and a half yeas and taught at National Chiao Tung Univesity, Taiwan, fo two yeas befoe joining The Chinese Univesity of Hong Kong in 982. He has published oiginal eseach on packet switched netwoks with contibutions in outing algoithms, buffe management, deadlock detection algoithms, message esequencing analysis, and multiaccess potocols. In ecent yeas, he banched out to wok on the design and analysis of cellula netwok, lightwave netwoks, and video distibution netwoks. He believes that the next challenge is designing an intelligent netwok that can accommodate the needs of individual customes. Pofesso Yum's eseach benefits a lot fom his gaduate students. Seven of them ae now pofessos at local and oveseas univesities. He is on the editoial boad of six jounals on communications and infomation science, including the ecently launched IEEE Tansactions on Multimedia. He is a senio membe of the IEEE.. Fo futhe infomation on this o any computing topic, please visit ou Digital Libay at

Slotted Random Access Protocol with Dynamic Transmission Probability Control in CDMA System

Slotted Random Access Protocol with Dynamic Transmission Probability Control in CDMA System Slotted Random Access Potocol with Dynamic Tansmission Pobability Contol in CDMA System Intaek Lim 1 1 Depatment of Embedded Softwae, Busan Univesity of Foeign Studies, itlim@bufs.ac.k Abstact In packet

More information

ANALYTIC PERFORMANCE MODELS FOR SINGLE CLASS AND MULTIPLE CLASS MULTITHREADED SOFTWARE SERVERS

ANALYTIC PERFORMANCE MODELS FOR SINGLE CLASS AND MULTIPLE CLASS MULTITHREADED SOFTWARE SERVERS ANALYTIC PERFORMANCE MODELS FOR SINGLE CLASS AND MULTIPLE CLASS MULTITHREADED SOFTWARE SERVERS Daniel A Menascé Mohamed N Bennani Dept of Compute Science Oacle, Inc Geoge Mason Univesity 1211 SW Fifth

More information

Any modern computer system will incorporate (at least) two levels of storage:

Any modern computer system will incorporate (at least) two levels of storage: 1 Any moden compute system will incopoate (at least) two levels of stoage: pimay stoage: andom access memoy (RAM) typical capacity 32MB to 1GB cost pe MB $3. typical access time 5ns to 6ns bust tansfe

More information

IP Network Design by Modified Branch Exchange Method

IP Network Design by Modified Branch Exchange Method Received: June 7, 207 98 IP Netwok Design by Modified Banch Method Kaiat Jaoenat Natchamol Sichumoenattana 2* Faculty of Engineeing at Kamphaeng Saen, Kasetsat Univesity, Thailand 2 Faculty of Management

More information

Journal of World s Electrical Engineering and Technology J. World. Elect. Eng. Tech. 1(1): 12-16, 2012

Journal of World s Electrical Engineering and Technology J. World. Elect. Eng. Tech. 1(1): 12-16, 2012 2011, Scienceline Publication www.science-line.com Jounal of Wold s Electical Engineeing and Technology J. Wold. Elect. Eng. Tech. 1(1): 12-16, 2012 JWEET An Efficient Algoithm fo Lip Segmentation in Colo

More information

IP Multicast Simulation in OPNET

IP Multicast Simulation in OPNET IP Multicast Simulation in OPNET Xin Wang, Chien-Ming Yu, Henning Schulzinne Paul A. Stipe Columbia Univesity Reutes Depatment of Compute Science 88 Pakway Dive South New Yok, New Yok Hauppuage, New Yok

More information

The Dual Round Robin Matching Switch with Exhaustive Service

The Dual Round Robin Matching Switch with Exhaustive Service The Dual Round Robin Matching Switch with Exhaustive Sevice Yihan Li, Shivenda S. Panwa, H. Jonathan Chao Abstact Vitual Output Queuing is widely used by fixed-length highspeed switches to ovecome head-of-line

More information

Point-Biserial Correlation Analysis of Fuzzy Attributes

Point-Biserial Correlation Analysis of Fuzzy Attributes Appl Math Inf Sci 6 No S pp 439S-444S (0 Applied Mathematics & Infomation Sciences An Intenational Jounal @ 0 NSP Natual Sciences Publishing o Point-iseial oelation Analysis of Fuzzy Attibutes Hao-En hueh

More information

A Memory Efficient Array Architecture for Real-Time Motion Estimation

A Memory Efficient Array Architecture for Real-Time Motion Estimation A Memoy Efficient Aay Achitectue fo Real-Time Motion Estimation Vasily G. Moshnyaga and Keikichi Tamau Depatment of Electonics & Communication, Kyoto Univesity Sakyo-ku, Yoshida-Honmachi, Kyoto 66-1, JAPAN

More information

Segmentation of Casting Defects in X-Ray Images Based on Fractal Dimension

Segmentation of Casting Defects in X-Ray Images Based on Fractal Dimension 17th Wold Confeence on Nondestuctive Testing, 25-28 Oct 2008, Shanghai, China Segmentation of Casting Defects in X-Ray Images Based on Factal Dimension Jue WANG 1, Xiaoqin HOU 2, Yufang CAI 3 ICT Reseach

More information

A New Finite Word-length Optimization Method Design for LDPC Decoder

A New Finite Word-length Optimization Method Design for LDPC Decoder A New Finite Wod-length Optimization Method Design fo LDPC Decode Jinlei Chen, Yan Zhang and Xu Wang Key Laboatoy of Netwok Oiented Intelligent Computation Shenzhen Gaduate School, Habin Institute of Technology

More information

RANDOM IRREGULAR BLOCK-HIERARCHICAL NETWORKS: ALGORITHMS FOR COMPUTATION OF MAIN PROPERTIES

RANDOM IRREGULAR BLOCK-HIERARCHICAL NETWORKS: ALGORITHMS FOR COMPUTATION OF MAIN PROPERTIES RANDOM IRREGULAR BLOCK-HIERARCHICAL NETWORKS: ALGORITHMS FOR COMPUTATION OF MAIN PROPERTIES Svetlana Avetisyan Mikayel Samvelyan* Matun Kaapetyan Yeevan State Univesity Abstact In this pape, the class

More information

CLOUD based storage systems are emerging to gain significant

CLOUD based storage systems are emerging to gain significant IEEE TRANSACTIONS ON CLOUD COMPUTING, VOL. 5, NO. 2, APRIL-JUNE 2017 221 On the Latency and Enegy Efficiency of Distibuted Stoage Systems Akshay Kuma, Student Membe, IEEE, Ravi Tandon, Membe, IEEE, and

More information

High performance CUDA based CNN image processor

High performance CUDA based CNN image processor High pefomance UDA based NN image pocesso GEORGE VALENTIN STOIA, RADU DOGARU, ELENA RISTINA STOIA Depatment of Applied Electonics and Infomation Engineeing Univesity Politehnica of Buchaest -3, Iuliu Maniu

More information

Achievable Catalog Size in Peer-to-Peer Video-on-Demand Systems

Achievable Catalog Size in Peer-to-Peer Video-on-Demand Systems Achievable Catalog Size in Pee-to-Pee Video-on-Demand Systems Yacine Boufkhad, Fabien Mathieu, Fabien De Montgolfie, Diego Peino, Lauent Viennot To cite this vesion: Yacine Boufkhad, Fabien Mathieu, Fabien

More information

An Unsupervised Segmentation Framework For Texture Image Queries

An Unsupervised Segmentation Framework For Texture Image Queries An Unsupevised Segmentation Famewok Fo Textue Image Queies Shu-Ching Chen Distibuted Multimedia Infomation System Laboatoy School of Compute Science Floida Intenational Univesity Miami, FL 33199, USA chens@cs.fiu.edu

More information

Optical Flow for Large Motion Using Gradient Technique

Optical Flow for Large Motion Using Gradient Technique SERBIAN JOURNAL OF ELECTRICAL ENGINEERING Vol. 3, No. 1, June 2006, 103-113 Optical Flow fo Lage Motion Using Gadient Technique Md. Moshaof Hossain Sake 1, Kamal Bechkoum 2, K.K. Islam 1 Abstact: In this

More information

Prioritized Traffic Recovery over GMPLS Networks

Prioritized Traffic Recovery over GMPLS Networks Pioitized Taffic Recovey ove GMPLS Netwoks 2005 IEEE. Pesonal use of this mateial is pemitted. Pemission fom IEEE mu be obtained fo all othe uses in any cuent o futue media including epinting/epublishing

More information

Modelling, simulation, and performance analysis of a CAN FD system with SAE benchmark based message set

Modelling, simulation, and performance analysis of a CAN FD system with SAE benchmark based message set Modelling, simulation, and pefomance analysis of a CAN FD system with SAE benchmak based message set Mahmut Tenuh, Panagiotis Oikonomidis, Peiklis Chachalakis, Elias Stipidis Mugla S. K. Univesity, TR;

More information

The Internet Ecosystem and Evolution

The Internet Ecosystem and Evolution The Intenet Ecosystem and Evolution Contents Netwok outing: basics distibuted/centalized, static/dynamic, linkstate/path-vecto inta-domain/inte-domain outing Mapping the sevice model to AS-AS paths valley-fee

More information

Exploring non-typical memcache architectures for decreased latency and distributed network usage.

Exploring non-typical memcache architectures for decreased latency and distributed network usage. Syacuse Univesity SURFACE Electical Engineeing and Compute Science Technical Repots College of Engineeing and Compute Science 9-5-20 Exploing non-typical memcache achitectues fo deceased latency and distibuted

More information

Controlled Information Maximization for SOM Knowledge Induced Learning

Controlled Information Maximization for SOM Knowledge Induced Learning 3 Int'l Conf. Atificial Intelligence ICAI'5 Contolled Infomation Maximization fo SOM Knowledge Induced Leaning Ryotao Kamimua IT Education Cente and Gaduate School of Science and Technology, Tokai Univeisity

More information

On using circuit-switched networks for file transfers

On using circuit-switched networks for file transfers On using cicuit-switched netwoks fo file tansfes Xiuduan Fang, Malathi Veeaaghavan Univesity of Viginia Email: {xf4c, mv5g}@viginia.edu Abstact High-speed optical cicuit-switched netwoks ae being deployed

More information

Communication vs Distributed Computation: an alternative trade-off curve

Communication vs Distributed Computation: an alternative trade-off curve Communication vs Distibuted Computation: an altenative tade-off cuve Yahya H. Ezzeldin, Mohammed amoose, Chistina Fagouli Univesity of Califonia, Los Angeles, CA 90095, USA, Email: {yahya.ezzeldin, mkamoose,

More information

Detection and Recognition of Alert Traffic Signs

Detection and Recognition of Alert Traffic Signs Detection and Recognition of Alet Taffic Signs Chia-Hsiung Chen, Macus Chen, and Tianshi Gao 1 Stanfod Univesity Stanfod, CA 9305 {echchen, macuscc, tianshig}@stanfod.edu Abstact Taffic signs povide dives

More information

SYSTEM LEVEL REUSE METRICS FOR OBJECT ORIENTED SOFTWARE : AN ALTERNATIVE APPROACH

SYSTEM LEVEL REUSE METRICS FOR OBJECT ORIENTED SOFTWARE : AN ALTERNATIVE APPROACH I J C A 7(), 202 pp. 49-53 SYSTEM LEVEL REUSE METRICS FOR OBJECT ORIENTED SOFTWARE : AN ALTERNATIVE APPROACH Sushil Goel and 2 Rajesh Vema Associate Pofesso, Depatment of Compute Science, Dyal Singh College,

More information

COSC 6385 Computer Architecture. - Pipelining

COSC 6385 Computer Architecture. - Pipelining COSC 6385 Compute Achitectue - Pipelining Sping 2012 Some of the slides ae based on a lectue by David Culle, Pipelining Pipelining is an implementation technique wheeby multiple instuctions ae ovelapped

More information

Lecture 8 Introduction to Pipelines Adapated from slides by David Patterson

Lecture 8 Introduction to Pipelines Adapated from slides by David Patterson Lectue 8 Intoduction to Pipelines Adapated fom slides by David Patteson http://www-inst.eecs.bekeley.edu/~cs61c/ * 1 Review (1/3) Datapath is the hadwae that pefoms opeations necessay to execute pogams.

More information

Combinatorial Mobile IP: A New Efficient Mobility Management Using Minimized Paging and Local Registration in Mobile IP Environments

Combinatorial Mobile IP: A New Efficient Mobility Management Using Minimized Paging and Local Registration in Mobile IP Environments Wieless Netwoks 0, 3 32, 200 200 Kluwe Academic Publishes. Manufactued in The Nethelands. Combinatoial Mobile IP: A New Efficient Mobility Management Using Minimized Paging and Local Registation in Mobile

More information

Modeling a shared medium access node with QoS distinction

Modeling a shared medium access node with QoS distinction Modeling a shaed medium access node with QoS distinction Matthias Gies, Jonas Geutet Compute Engineeing and Netwoks Laboatoy (TIK) Swiss Fedeal Institute of Technology Züich CH-8092 Züich, Switzeland email:

More information

Number of Paths and Neighbours Effect on Multipath Routing in Mobile Ad Hoc Networks

Number of Paths and Neighbours Effect on Multipath Routing in Mobile Ad Hoc Networks Numbe of Paths and Neighbous Effect on Multipath Routing in Mobile Ad Hoc Netwoks Oday Jeew School of Infomation Systems and Accounting Univesity of Canbea Canbea ACT 2617, Austalia oday.jeew@canbea.edu.au

More information

Method of controlling access to intellectual switching nodes of telecommunication networks and systems

Method of controlling access to intellectual switching nodes of telecommunication networks and systems ISSN (e): 2250 3005 Volume 05 Issue 05 ay 2015 Intenational Jounal of Computational Engineeing eseach (IJCE) ethod of contolling access to intellectual switching nodes of telecommunication netwoks and

More information

Illumination methods for optical wear detection

Illumination methods for optical wear detection Illumination methods fo optical wea detection 1 J. Zhang, 2 P.P.L.Regtien 1 VIMEC Applied Vision Technology, Coy 43, 5653 LC Eindhoven, The Nethelands Email: jianbo.zhang@gmail.com 2 Faculty Electical

More information

Accurate Diffraction Efficiency Control for Multiplexed Volume Holographic Gratings. Xuliang Han, Gicherl Kim, and Ray T. Chen

Accurate Diffraction Efficiency Control for Multiplexed Volume Holographic Gratings. Xuliang Han, Gicherl Kim, and Ray T. Chen Accuate Diffaction Efficiency Contol fo Multiplexed Volume Hologaphic Gatings Xuliang Han, Gichel Kim, and Ray T. Chen Micoelectonic Reseach Cente Depatment of Electical and Compute Engineeing Univesity

More information

Gravitational Shift for Beginners

Gravitational Shift for Beginners Gavitational Shift fo Beginnes This pape, which I wote in 26, fomulates the equations fo gavitational shifts fom the elativistic famewok of special elativity. Fist I deive the fomulas fo the gavitational

More information

WIRELESS sensor networks (WSNs), which are capable

WIRELESS sensor networks (WSNs), which are capable IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. XX, NO. XX, XXX 214 1 Lifetime and Enegy Hole Evolution Analysis in Data-Gatheing Wieless Senso Netwoks Ju Ren, Student Membe, IEEE, Yaoxue Zhang, Kuan

More information

ANN Models for Coplanar Strip Line Analysis and Synthesis

ANN Models for Coplanar Strip Line Analysis and Synthesis 200 IJCSNS Intenational Jounal of Compute Science and Netwok Secuity, VOL.8 No.10, Octobe 2008 Models fo Coplana Stip Line Analysis and J.Lakshmi Naayana D.K.Si Rama Kishna D.L.Patap Reddy Chalapathi Institute

More information

HISTOGRAMS are an important statistic reflecting the

HISTOGRAMS are an important statistic reflecting the JOURNAL OF L A T E X CLASS FILES, VOL. 14, NO. 8, AUGUST 2015 1 D 2 HistoSketch: Disciminative and Dynamic Similaity-Peseving Sketching of Steaming Histogams Dingqi Yang, Bin Li, Laua Rettig, and Philippe

More information

Frequency Domain Approach for Face Recognition Using Optical Vanderlugt Filters

Frequency Domain Approach for Face Recognition Using Optical Vanderlugt Filters Optics and Photonics Jounal, 016, 6, 94-100 Published Online August 016 in SciRes. http://www.scip.og/jounal/opj http://dx.doi.og/10.436/opj.016.68b016 Fequency Domain Appoach fo Face Recognition Using

More information

Lecture Topics ECE 341. Lecture # 12. Control Signals. Control Signals for Datapath. Basic Processing Unit. Pipelining

Lecture Topics ECE 341. Lecture # 12. Control Signals. Control Signals for Datapath. Basic Processing Unit. Pipelining EE 341 Lectue # 12 Instucto: Zeshan hishti zeshan@ece.pdx.edu Novembe 10, 2014 Potland State Univesity asic Pocessing Unit ontol Signals Hadwied ontol Datapath contol signals Dealing with memoy delay Pipelining

More information

Concomitants of Upper Record Statistics for Bivariate Pseudo Weibull Distribution

Concomitants of Upper Record Statistics for Bivariate Pseudo Weibull Distribution Available at http://pvamuedu/aam Appl Appl Math ISSN: 93-9466 Vol 5, Issue (Decembe ), pp 8 9 (Peviously, Vol 5, Issue, pp 379 388) Applications and Applied Mathematics: An Intenational Jounal (AAM) Concomitants

More information

dc - Linux Command Dc may be invoked with the following command-line options: -V --version Print out the version of dc

dc - Linux Command Dc may be invoked with the following command-line options: -V --version Print out the version of dc - CentOS 5.2 - Linux Uses Guide - Linux Command SYNOPSIS [-V] [--vesion] [-h] [--help] [-e sciptexpession] [--expession=sciptexpession] [-f sciptfile] [--file=sciptfile] [file...] DESCRIPTION is a evese-polish

More information

Image Enhancement in the Spatial Domain. Spatial Domain

Image Enhancement in the Spatial Domain. Spatial Domain 8-- Spatial Domain Image Enhancement in the Spatial Domain What is spatial domain The space whee all pixels fom an image In spatial domain we can epesent an image by f( whee x and y ae coodinates along

More information

Spiral Recognition Methodology and Its Application for Recognition of Chinese Bank Checks

Spiral Recognition Methodology and Its Application for Recognition of Chinese Bank Checks Spial Recognition Methodology and Its Application fo Recognition of Chinese Bank Checks Hanshen Tang 1, Emmanuel Augustin 2, Ching Y. Suen 1, Olivie Baet 2, Mohamed Cheiet 3 1 Cente fo Patten Recognition

More information

SCALABLE ENERGY EFFICIENT AD-HOC ON DEMAND DISTANCE VECTOR (SEE-AODV) ROUTING PROTOCOL IN WIRELESS MESH NETWORKS

SCALABLE ENERGY EFFICIENT AD-HOC ON DEMAND DISTANCE VECTOR (SEE-AODV) ROUTING PROTOCOL IN WIRELESS MESH NETWORKS SCALABL NRGY FFICINT AD-HOC ON DMAND DISTANC VCTOR (S-AODV) ROUTING PROTOCOL IN WIRLSS MSH NTWORKS Sikande Singh Reseach Schola, Depatment of Compute Science & ngineeing, Punjab ngineeing College (PC),

More information

arxiv: v2 [physics.soc-ph] 30 Nov 2016

arxiv: v2 [physics.soc-ph] 30 Nov 2016 Tanspotation dynamics on coupled netwoks with limited bandwidth Ming Li 1,*, Mao-Bin Hu 1, and Bing-Hong Wang 2, axiv:1607.05382v2 [physics.soc-ph] 30 Nov 2016 1 School of Engineeing Science, Univesity

More information

Worst-Case Delay Bounds for Uniform Load-Balanced Switch Fabrics

Worst-Case Delay Bounds for Uniform Load-Balanced Switch Fabrics Wost-Case Delay Bounds fo Unifom Load-Balanced Switch Fabics Spyidon Antonakopoulos, Steven Fotune, Rae McLellan, Lisa Zhang Bell Laboatoies, 600 Mountain Ave, Muay Hill, NJ 07974 fistname.lastname@alcatel-lucent.com

More information

An Efficient Handover Mechanism Using the General Switch Management Protocol on a Multi-Protocol Label Switching Network

An Efficient Handover Mechanism Using the General Switch Management Protocol on a Multi-Protocol Label Switching Network An Efficient andove Mechanism Using the Geneal Switch Management Potocol on a Multi-Potocol abel Switching Netwok Seong Gon hoi, yun Joo Kang, and Jun Kyun hoi Using the geneal switch management potocol

More information

On Adaptive Bandwidth Sharing with Rate Guarantees

On Adaptive Bandwidth Sharing with Rate Guarantees On Adaptive Bandwidth Shaing with Rate Guaantees N.G. Duffield y T. V. Lakshman D. Stiliadis y AT&T Laboatoies Bell Labs Rm A175, 180 Pak Avenue Lucent Technologies Floham Pak, 101 Cawfods Cone Road NJ

More information

On Error Estimation in Runge-Kutta Methods

On Error Estimation in Runge-Kutta Methods Leonado Jounal of Sciences ISSN 1583-0233 Issue 18, Januay-June 2011 p. 1-10 On Eo Estimation in Runge-Kutta Methods Ochoche ABRAHAM 1,*, Gbolahan BOLARIN 2 1 Depatment of Infomation Technology, 2 Depatment

More information

Fifth Wheel Modelling and Testing

Fifth Wheel Modelling and Testing Fifth heel Modelling and Testing en Masoy Mechanical Engineeing Depatment Floida Atlantic Univesity Boca aton, FL 4 Lois Malaptias IFMA Institut Fancais De Mechanique Advancee ampus De lemont Feand Les

More information

Title. Author(s)NOMURA, K.; MOROOKA, S. Issue Date Doc URL. Type. Note. File Information

Title. Author(s)NOMURA, K.; MOROOKA, S. Issue Date Doc URL. Type. Note. File Information Title CALCULATION FORMULA FOR A MAXIMUM BENDING MOMENT AND THE TRIANGULAR SLAB WITH CONSIDERING EFFECT OF SUPPO UNIFORM LOAD Autho(s)NOMURA, K.; MOROOKA, S. Issue Date 2013-09-11 Doc URL http://hdl.handle.net/2115/54220

More information

On the Forwarding Area of Contention-Based Geographic Forwarding for Ad Hoc and Sensor Networks

On the Forwarding Area of Contention-Based Geographic Forwarding for Ad Hoc and Sensor Networks On the Fowading Aea of Contention-Based Geogaphic Fowading fo Ad Hoc and Senso Netwoks Dazhi Chen Depatment of EECS Syacuse Univesity Syacuse, NY dchen@sy.edu Jing Deng Depatment of CS Univesity of New

More information

(1) W tcp = (3) N. Assuming 1 P r 1. = W r (4) a 1/(k+1) W 2/(k+1)

(1) W tcp = (3) N. Assuming 1 P r 1. = W r (4) a 1/(k+1) W 2/(k+1) 1 Multi Path PERT Ankit Singh and A. L. Naasimha Reddy Electical and Compute Engineeing Depatment, Texas A&M Univesity; email: eddy@ece.tamu.edu. Abstact This pape pesents a new multipath delay based algoithm,

More information

Delay-Optimized File Retrieval under LT-Based Cloud Storage

Delay-Optimized File Retrieval under LT-Based Cloud Storage infomation: DOI 10.1109/TCC.015.404, IEEE Tansactions on Cloud Computing IEEE TRANSACTIONS ON CLOUD COMPUTING 1 Delay-Optimized File Retieval unde LT-Based Cloud Stoage Haifeng Lu,Chuan Heng Foh, Senio

More information

Cellular Neural Network Based PTV

Cellular Neural Network Based PTV 3th Int Symp on Applications of Lase Techniques to Fluid Mechanics Lisbon, Potugal, 6-9 June, 006 Cellula Neual Netwok Based PT Kazuo Ohmi, Achyut Sapkota : Depatment of Infomation Systems Engineeing,

More information

Improvement of First-order Takagi-Sugeno Models Using Local Uniform B-splines 1

Improvement of First-order Takagi-Sugeno Models Using Local Uniform B-splines 1 Impovement of Fist-ode Takagi-Sugeno Models Using Local Unifom B-splines Felipe Fenández, Julio Gutiéez, Gacián Tiviño and Juan Calos Cespo Dep. Tecnología Fotónica, Facultad de Infomática Univesidad Politécnica

More information

Obstacle Avoidance of Autonomous Mobile Robot using Stereo Vision Sensor

Obstacle Avoidance of Autonomous Mobile Robot using Stereo Vision Sensor Obstacle Avoidance of Autonomous Mobile Robot using Steeo Vision Senso Masako Kumano Akihisa Ohya Shin ichi Yuta Intelligent Robot Laboatoy Univesity of Tsukuba, Ibaaki, 35-8573 Japan E-mail: {masako,

More information

Module 6 STILL IMAGE COMPRESSION STANDARDS

Module 6 STILL IMAGE COMPRESSION STANDARDS Module 6 STILL IMAE COMPRESSION STANDARDS Lesson 17 JPE-2000 Achitectue and Featues Instuctional Objectives At the end of this lesson, the students should be able to: 1. State the shotcomings of JPE standad.

More information

Dynamic Topology Control to Reduce Interference in MANETs

Dynamic Topology Control to Reduce Interference in MANETs Dynamic Topology Contol to Reduce Intefeence in MANETs Hwee Xian TAN 1,2 and Winston K. G. SEAH 2,1 {stuhxt, winston}@i2.a-sta.edu.sg 1 Depatment of Compute Science, School of Computing, National Univesity

More information

Fault-Tolerant Routing Schemes in RDT(2,2,1)/α-Based Interconnection Network for Networks-on-Chip Designs

Fault-Tolerant Routing Schemes in RDT(2,2,1)/α-Based Interconnection Network for Networks-on-Chip Designs Fault-Toleant Routing Schemes in RDT(,,)/α-Based Inteconnection Netwok fo Netwoks-on-Chip Designs Mei Yang, Tao Li, Yingtao Jiang, and Yulu Yang Dept. of Electical & Compute Engineeing Univesity of Nevada,

More information

EE 6900: Interconnection Networks for HPC Systems Fall 2016

EE 6900: Interconnection Networks for HPC Systems Fall 2016 EE 6900: Inteconnection Netwoks fo HPC Systems Fall 2016 Avinash Kaanth Kodi School of Electical Engineeing and Compute Science Ohio Univesity Athens, OH 45701 Email: kodi@ohio.edu 1 Acknowledgement: Inteconnection

More information

A modal estimation based multitype sensor placement method

A modal estimation based multitype sensor placement method A modal estimation based multitype senso placement method *Xue-Yang Pei 1), Ting-Hua Yi 2) and Hong-Nan Li 3) 1),)2),3) School of Civil Engineeing, Dalian Univesity of Technology, Dalian 116023, China;

More information

A Two-stage and Parameter-free Binarization Method for Degraded Document Images

A Two-stage and Parameter-free Binarization Method for Degraded Document Images A Two-stage and Paamete-fee Binaization Method fo Degaded Document Images Yung-Hsiang Chiu 1, Kuo-Liang Chung 1, Yong-Huai Huang 2, Wei-Ning Yang 3, Chi-Huang Liao 4 1 Depatment of Compute Science and

More information

Positioning of a robot based on binocular vision for hand / foot fusion Long Han

Positioning of a robot based on binocular vision for hand / foot fusion Long Han 2nd Intenational Confeence on Advances in Mechanical Engineeing and Industial Infomatics (AMEII 26) Positioning of a obot based on binocula vision fo hand / foot fusion Long Han Compute Science and Technology,

More information

= dv 3V (r + a 1) 3 r 3 f(r) = 1. = ( (r + r 2

= dv 3V (r + a 1) 3 r 3 f(r) = 1. = ( (r + r 2 Random Waypoint Model in n-dimensional Space Esa Hyytiä and Joma Vitamo Netwoking Laboatoy, Helsinki Univesity of Technology, Finland Abstact The andom waypoint model (RWP) is one of the most widely used

More information

DEADLOCK AVOIDANCE IN BATCH PROCESSES. M. Tittus K. Åkesson

DEADLOCK AVOIDANCE IN BATCH PROCESSES. M. Tittus K. Åkesson DEADLOCK AVOIDANCE IN BATCH PROCESSES M. Tittus K. Åkesson Univesity College Boås, Sweden, e-mail: Michael.Tittus@hb.se Chalmes Univesity of Technology, Gothenbug, Sweden, e-mail: ka@s2.chalmes.se Abstact:

More information

Shortest Paths for a Two-Robot Rendez-Vous

Shortest Paths for a Two-Robot Rendez-Vous Shotest Paths fo a Two-Robot Rendez-Vous Eik L Wyntes Joseph S B Mitchell y Abstact In this pape, we conside an optimal motion planning poblem fo a pai of point obots in a plana envionment with polygonal

More information

Configuring RSVP-ATM QoS Interworking

Configuring RSVP-ATM QoS Interworking Configuing RSVP-ATM QoS Intewoking Last Updated: Januay 15, 2013 This chapte descibes the tasks fo configuing the RSVP-ATM QoS Intewoking featue, which povides suppot fo Contolled Load Sevice using RSVP

More information

Lecture #22 Pipelining II, Cache I

Lecture #22 Pipelining II, Cache I inst.eecs.bekeley.edu/~cs61c CS61C : Machine Stuctues Lectue #22 Pipelining II, Cache I Wiewold cicuits 2008-7-29 http://www.maa.og/editoial/mathgames/mathgames_05_24_04.html http://www.quinapalus.com/wi-index.html

More information

Adaptation of TDMA Parameters Based on Network Conditions

Adaptation of TDMA Parameters Based on Network Conditions Adaptation of TDMA Paametes Based on Netwok Conditions Boa Kaaoglu Dept. of Elect. and Compute Eng. Univesity of Rocheste Rocheste, NY 14627 Email: kaaoglu@ece.ocheste.edu Tolga Numanoglu Dept. of Elect.

More information

Conversion Functions for Symmetric Key Ciphers

Conversion Functions for Symmetric Key Ciphers Jounal of Infomation Assuance and Secuity 2 (2006) 41 50 Convesion Functions fo Symmetic Key Ciphes Deba L. Cook and Angelos D. Keomytis Depatment of Compute Science Columbia Univesity, mail code 0401

More information

A Novel Automatic White Balance Method For Digital Still Cameras

A Novel Automatic White Balance Method For Digital Still Cameras A Novel Automatic White Balance Method Fo Digital Still Cameas Ching-Chih Weng 1, Home Chen 1,2, and Chiou-Shann Fuh 3 Depatment of Electical Engineeing, 2 3 Gaduate Institute of Communication Engineeing

More information

MapReduce Optimizations and Algorithms 2015 Professor Sasu Tarkoma

MapReduce Optimizations and Algorithms 2015 Professor Sasu Tarkoma apreduce Optimizations and Algoithms 2015 Pofesso Sasu Takoma www.cs.helsinki.fi Optimizations Reduce tasks cannot stat befoe the whole map phase is complete Thus single slow machine can slow down the

More information

Simulation and Performance Evaluation of Network on Chip Architectures and Algorithms using CINSIM

Simulation and Performance Evaluation of Network on Chip Architectures and Algorithms using CINSIM J. Basic. Appl. Sci. Res., 1(10)1594-1602, 2011 2011, TextRoad Publication ISSN 2090-424X Jounal of Basic and Applied Scientific Reseach www.textoad.com Simulation and Pefomance Evaluation of Netwok on

More information

Monte Carlo Simulation for the ECAT HRRT using GATE

Monte Carlo Simulation for the ECAT HRRT using GATE Monte Calo Simulation fo the ECAT HRRT using GATE F. Bataille, C. Comtat, Membe, IEEE, S. Jan, and R. Tébossen Abstact The ECAT HRRT (High Resolution Reseach Tomogaph, CPS Innovations, Knoxville, TN, U.S.A.)

More information

INFORMATION DISSEMINATION DELAY IN VEHICLE-TO-VEHICLE COMMUNICATION NETWORKS IN A TRAFFIC STREAM

INFORMATION DISSEMINATION DELAY IN VEHICLE-TO-VEHICLE COMMUNICATION NETWORKS IN A TRAFFIC STREAM INFORMATION DISSEMINATION DELAY IN VEHICLE-TO-VEHICLE COMMUNICATION NETWORKS IN A TRAFFIC STREAM LiLi Du Depatment of Civil, Achitectual, and Envionmental Engineeing Illinois Institute of Technology 3300

More information

Towards Adaptive Information Merging Using Selected XML Fragments

Towards Adaptive Information Merging Using Selected XML Fragments Towads Adaptive Infomation Meging Using Selected XML Fagments Ho-Lam Lau and Wilfed Ng Depatment of Compute Science and Engineeing, The Hong Kong Univesity of Science and Technology, Hong Kong {lauhl,

More information

Experimental and numerical simulation of the flow over a spillway

Experimental and numerical simulation of the flow over a spillway Euopean Wate 57: 253-260, 2017. 2017 E.W. Publications Expeimental and numeical simulation of the flow ove a spillway A. Seafeim *, L. Avgeis, V. Hissanthou and K. Bellos Depatment of Civil Engineeing,

More information

Introduction To Pipelining. Chapter Pipelining1 1

Introduction To Pipelining. Chapter Pipelining1 1 Intoduction To Pipelining Chapte 6.1 - Pipelining1 1 Mooe s Law Mooe s Law says that the numbe of pocessos on a chip doubles about evey 18 months. Given the data on the following two slides, is this tue?

More information

An Improved Resource Reservation Protocol

An Improved Resource Reservation Protocol Jounal of Compute Science 3 (8: 658-665, 2007 SSN 549-3636 2007 Science Publications An mpoved Resouce Resevation Potocol Desie Oulai, Steven Chambeland and Samuel Piee Depatment of Compute Engineeing

More information

Reachable State Spaces of Distributed Deadlock Avoidance Protocols

Reachable State Spaces of Distributed Deadlock Avoidance Protocols Reachable State Spaces of Distibuted Deadlock Avoidance Potocols CÉSAR SÁNCHEZ and HENNY B. SIPMA Stanfod Univesity We pesent a family of efficient distibuted deadlock avoidance algoithms with applications

More information

Erasure-Coding Based Routing for Opportunistic Networks

Erasure-Coding Based Routing for Opportunistic Networks Easue-Coding Based Routing fo Oppotunistic Netwoks Yong Wang, Sushant Jain, Magaet Matonosi, Kevin Fall Pinceton Univesity, Univesity of Washington, Intel Reseach Bekeley ABSTRACT Routing in Delay Toleant

More information

Annales UMCS Informatica AI 2 (2004) UMCS

Annales UMCS Informatica AI 2 (2004) UMCS Pobane z czasopisma Annales AI- Infomatica http://ai.annales.umcs.pl Annales Infomatica AI 2 (2004) 33-340 Annales Infomatica Lublin-Polonia Sectio AI http://www.annales.umcs.lublin.pl/ Embedding as a

More information

Accelerating Storage with RDMA Max Gurtovoy Mellanox Technologies

Accelerating Storage with RDMA Max Gurtovoy Mellanox Technologies Acceleating Stoage with RDMA Max Gutovoy Mellanox Technologies 2018 Stoage Develope Confeence EMEA. Mellanox Technologies. All Rights Reseved. 1 What is RDMA? Remote Diect Memoy Access - povides the ability

More information

XFVHDL: A Tool for the Synthesis of Fuzzy Logic Controllers

XFVHDL: A Tool for the Synthesis of Fuzzy Logic Controllers XFVHDL: A Tool fo the Synthesis of Fuzzy Logic Contolles E. Lago, C. J. Jiménez, D. R. López, S. Sánchez-Solano and A. Baiga Instituto de Micoelectónica de Sevilla. Cento Nacional de Micoelectónica, Edificio

More information

Performance Optimization in Structured Wireless Sensor Networks

Performance Optimization in Structured Wireless Sensor Networks 5 The Intenational Aab Jounal of Infomation Technology, Vol. 6, o. 5, ovembe 9 Pefomance Optimization in Stuctued Wieless Senso etwoks Amine Moussa and Hoda Maalouf Compute Science Depatment, ote Dame

More information

On the Conversion between Binary Code and Binary-Reflected Gray Code on Boolean Cubes

On the Conversion between Binary Code and Binary-Reflected Gray Code on Boolean Cubes On the Convesion between Binay Code and BinayReflected Gay Code on Boolean Cubes The Havad community has made this aticle openly available. Please shae how this access benefits you. You stoy mattes Citation

More information

Separability and Topology Control of Quasi Unit Disk Graphs

Separability and Topology Control of Quasi Unit Disk Graphs Sepaability and Topology Contol of Quasi Unit Disk Gaphs Jiane Chen, Anxiao(Andew) Jiang, Iyad A. Kanj, Ge Xia, and Fenghui Zhang Dept. of Compute Science, Texas A&M Univ. College Station, TX 7784. {chen,

More information

Color Correction Using 3D Multiview Geometry

Color Correction Using 3D Multiview Geometry Colo Coection Using 3D Multiview Geomety Dong-Won Shin and Yo-Sung Ho Gwangju Institute of Science and Technology (GIST) 13 Cheomdan-gwagio, Buk-ku, Gwangju 500-71, Republic of Koea ABSTRACT Recently,

More information

5 4 THE BERNOULLI EQUATION

5 4 THE BERNOULLI EQUATION 185 CHATER 5 the suounding ai). The fictional wok tem w fiction is often expessed as e loss to epesent the loss (convesion) of mechanical into themal. Fo the idealied case of fictionless motion, the last

More information

A Family of Distributed Deadlock Avoidance Protocols and their Reachable State Spaces

A Family of Distributed Deadlock Avoidance Protocols and their Reachable State Spaces A Family of Distibuted Deadlock Avoidance Potocols and thei Reachable State Spaces Césa Sánchez, Henny B. Sipma, and Zoha Manna Compute Science Depatment Stanfod Univesity, Stanfod, CA 94305-9025 {cesa,sipma,manna}@cs.stanfod.edu

More information

A Shape-preserving Affine Takagi-Sugeno Model Based on a Piecewise Constant Nonuniform Fuzzification Transform

A Shape-preserving Affine Takagi-Sugeno Model Based on a Piecewise Constant Nonuniform Fuzzification Transform A Shape-peseving Affine Takagi-Sugeno Model Based on a Piecewise Constant Nonunifom Fuzzification Tansfom Felipe Fenández, Julio Gutiéez, Juan Calos Cespo and Gacián Tiviño Dep. Tecnología Fotónica, Facultad

More information

Cryptanalysis of Hwang-Chang s a Time-Stamp Protocol for Digital Watermarking

Cryptanalysis of Hwang-Chang s a Time-Stamp Protocol for Digital Watermarking Cyptanalysis of Hwang-Chang s a Time-Stamp Potocol fo Digital Watemaking *Jue-Sam Chou, Yalin Chen 2, Chung-Ju Chan 3 Depatment of Infomation Management, Nanhua Univesity Chiayi 622 Taiwan, R.O.C *: coesponding

More information

Scaling Location-based Services with Dynamically Composed Location Index

Scaling Location-based Services with Dynamically Composed Location Index Scaling Location-based Sevices with Dynamically Composed Location Index Bhuvan Bamba, Sangeetha Seshadi and Ling Liu Distibuted Data Intensive Systems Laboatoy (DiSL) College of Computing, Geogia Institute

More information

ADDING REALISM TO SOURCE CHARACTERIZATION USING A GENETIC ALGORITHM

ADDING REALISM TO SOURCE CHARACTERIZATION USING A GENETIC ALGORITHM ADDING REALISM TO SOURCE CHARACTERIZATION USING A GENETIC ALGORITHM Luna M. Rodiguez*, Sue Ellen Haupt, and Geoge S. Young Depatment of Meteoology and Applied Reseach Laboatoy The Pennsylvania State Univesity,

More information

Journal of Network and Computer Applications

Journal of Network and Computer Applications Jounal of Netwok and Compute Applications 34 (211) 135 142 Contents lists available at ScienceDiect Jounal of Netwok and Compute Applications jounal homepage: www.elsevie.com/locate/jnca Optimization of

More information

Computer Science 141 Computing Hardware

Computer Science 141 Computing Hardware Compute Science 141 Computing Hadwae Fall 2006 Havad Univesity Instucto: Pof. David Books dbooks@eecs.havad.edu [MIPS Pipeline Slides adapted fom Dave Patteson s UCB CS152 slides and May Jane Iwin s CSE331/431

More information

Cloud-Aided Wireless Networks with Edge Caching: Fundamental Latency Trade-Offs in Fog Radio Access Networks

Cloud-Aided Wireless Networks with Edge Caching: Fundamental Latency Trade-Offs in Fog Radio Access Networks 206 IEEE Intenational Symposium on Infomation Theoy Cloud-Aided Wieless Netwos with Edge Caching: Fundamental Latency Tade-Offs in Fog Radio Access Netwos Ravi Tandon Osvaldo Simeone Depatment of ECE Univesity

More information

THE THETA BLOCKCHAIN

THE THETA BLOCKCHAIN THE THETA BLOCKCHAIN Theta is a decentalized video steaming netwok, poweed by a new blockchain and token. By Theta Labs, Inc. Last Updated: Nov 21, 2017 esion 1.0 1 OUTLINE Motivation Reputation Dependent

More information