TCP Timeout Mechanism for Optimization of Network Fairness and Performance in Multi-Hop Pipeline Network

Transcription

1 Draft Version - Brunel University, Lonon TCP Timeout Mechanism for Optimization of Network Fairness an Performance in Multi-Hop Pipeline Network Siva Kumar Subramaniam 1, Rajagopal Nilavalan 3, Wamaeva Balachanran 4 Department of Electronic & Computer Eng., College of Engineering, Design & Physical Sciences, Brunel University Lonon, Unite Kingom Siva@brunel.ac.uk Shariq Mahmoo Khan 2 Department of Computer Science & IT, NED University of Engineering & Technology, Pakistan shariq@neuet.eu.pk Abstract In the recent years, wireless sensor network (WSN) has a huge impact in many remote base applications especially in oil an gas pipeline monitoring. Thus, the eployment of a multi-hop linear WSN will be a practical solution on pipelines. With a large multi-hop linear WSN, the overall network performance is baly affecte especially ue to noe starvation. Inequality among source noes is relatively an amplifie factor of generate ata rate an source noe istances from the estination noe. In this paper, we propose a mathematical moel for TCP Delaye Timeout Acknowlegement (DTO-ACK) mechanism for non-zero passive noes. Optimum throughput fairness can be achieve by implementation of DTO-ACK with Dual Interleaving Linear Static Routing Protocol (DI-LSRP) for flat topology. Implementation of DTO-ACK an moification of TCP parameters reuces packet collision an ensures optimal throughput fairness in a multi-hop linear topology using TCP agent. Keywors TCP, throughput fairness inex, elaye acknowlegement, linear topology, pipeline network I. INTRODUCTION The eman for remote monitoring is establishe with various communication technologies in relaying real-time ata to a centralise monitoring station from sensors attache to critical pipeline points [1-3]. In a static geographical infrastructure of an oil an gas pipeline usually stretche over hunres of miles between the sensing points (source noes) to a monitoring stations (ata collection for future analysis). The network of sensors which are linke in series to form a chain to sense changes on sensing points an transfer this information to a estination noe with preefine route etermine by a routing algorithm. Such a pipeline network has unique unerlying issues effecting full eployment of WSN on pipelines in terms of performance as well as fairness among noes [4-7]. In a multihop WSN, the long range ata transmission [1] has severe performance egraation issues which lea towars inefficient fairness in the network particularly using the Transmission Control Protocol (TCP) [4, 7, 8]. In the recent years, continuous research on new protocols, applications, techniques an evices have been stuie in TCP for achieving optimum fairness outcomes. TCP experience performance issues where the root cause is from noe starvation an unbalance ata packet flow in a long range linear topology. II. RELATED WORKS TCP is a popular agent in many IEEE base application where various mechanism with high performance an congestion collapse avoiance is achieve [9, 10]. The key function of a TCP mechanism is to coorinate the ata sening rate in a WSN at an optimum state. Avertise winow at the estination noe an congestion winow at the source noes overseas the banwith utilisation rate for the purposes of ata packet transmission. The newer version of TCP mechanism is built with multiple algorithms such as congestion avoiance, fast retransmit, slow-start an fast recovery as state in (RFC 5681) [11]. TCP mechanism at a source noe employs a retransmission time-out (RTO) mechanism for estimate Roun Trip Time (RTT). RTT is one cycle time from a source noe to sen a ata packet to a estination noe an to receive an acknowlegement packet in return to the originator. The etaile characteristic of RTT is as state in RFC 2988 [12]. Many types of research are conucte in enhancing TCP [8] performance in hanling congestion, reucing errors, hanling loss an etc. hence there are many types of TCP algorithms available for users. Delaye cumulative acknowlegement (TCP-DCA) [13] is a scheme with variable elay winow size which aapts to the TCP acknowlegement packet generation base on hop counts. Path length in proportional to the istance is an important metrics in eciing the elay winow size for the number of acknowlegement packets to be returne to a source noe. Cooperation between channel access control along with TCP Rate Aaptation (CATRA) [14] will etermine an optimal Contention Winow (CW) size to the control station contention. The moifie CW size for achieving per-station fairness an the channel utilisation information is sent to the transport layer for TCP sening rate ajustment to achieve per-flow fairness. TCP- Vegas-W [15] is a metho use to reuce aggressive winow size in a multi-hop wireless network. With this metho, the congestion winow can be operate at an appropriate level. TCP

2 Draft Version - Brunel University, Lonon with Aaptive Delay Winow (TCP-ADW) [16] is esigne to reuce the acknowlegement packets appropriately by reaching an optimal ynamic elay winow. The elay avois the transmission time-out at the source noes unless the retransmission timer expires, the elay winow will be increase by the estination noe base on the ata rate. Aaptive Delaye ACK Mechanism (ADAM) [7] works by properly ajusting TCP parameter which affects the throughput fairness an further to reuce frame collisions. ADAM+ [7] is another metho introuce by the author to improve throughput fairness further by eliminating RTS/CTS. Monitoring Delaye Acknowlegment (TCP-MDA) [17] is esigne to interact with TCP an MAC layer to reuce the acknowlegement packets by channel conition monitoring. III. TCP DELAYED TIMEOUT ACKNOWLEDGEMENT (DTO-ACK) The propose TCP Delaye Timeout Acknowlegement (DTO-ACK) moel is formulate for TCP agent acknowlegement timeout in a multi-hop linear topology. The propose moel woul be a practical solution in an application where the normalise throughput fairness is esire at a ratio of one between all source noes in the network. To achieve the ratio of (1:1) among all source noes, the source noe winow size (w) equals to min value of (awn, cwn), where avertise winow is (awn) an congestion winow is (cwn). n hops 1 hop Sequence no.1 Sequence no.2 Sequence no.3 Sequence no.4 En On E1 O1 n hops Ack-O1 avertise winow is set to one (the lowest value) an the congestion winow is ecie by the TCP congestion control algorithm base the congestion state in the network. The efault elaye acknowlegement factor of two [7, 9] is controlle by setting the acknowlegement time after the efault acknowlegement expiry time thus retains the fairness inex value close to one. The Dual Interleaving Linear Static Routing Protocol (DI-LSRP) with DTO-ACK mathematical formulation is use to set an ieal elaye acknowlegement time for each source noes in a multi-hop linear topology as shown in Fig. 1. Where On/En is the maximum number of source noes in each path with a maximum communication range in istance of 2. DILSRP is a static prefix ual path (O/Even) routing algorithm in a linear arrangement to a estination noe. Generally, in TCP agents elaye acknowlegement factor is set at two (RFC1122) [7, 9] where an acknowlegement packet will be generate when the estination noe receives one or two ata packets within the efault acknowlegement expiry time. The RTT is one cycle time taken from a source noe to sen a ata packet to the estination noe an receive an acknowlegement packet to confirm its recipient of the sent ata packet. In DTO-ACK moel, RTT is a multiplying factor with the total number of hops for all source noes in a certain network. A total number of hops (travel cost) is a very important parameter to etermine appropriate elaye acknowlegement time for each source noes as formulate in (1-2). The RTT value is use in calculating an ieal elaye acknowlegement time for each respective source noes in the network base on istance (calculate by the total hops neee to reach a estination noe). Sum H = + (1) Where Sum H is the total hop count for n number of noes, HiE is the total hop count for n number of noes (Even) an HiO is the total hop count for n number of noes (O) ND Sequence no.1 Sequence no.2 Sequence no.3 Sequence no.4 1 hop Fig. 1: The DTO-ACK moel with On/En number of source noes an ND as estination noe evenly istribute in within half the communication range. The avertise winow is esigne to restrict the operation of source noes within the available resources at the estination noe whereas the congestion winow is esigne to avoi the source noes from over sening ata packets than the permitte allocations in a certain network. To achieve the esirable normalise throughput fairness among all source noes, the Fair cycle time (t) = Sum H RTT (2) Where Fair cycle time (t) is one cycle time taken for n number of noes in the network to have a fair ata transmission to the estination noe, Sum H is from (2), RTT is the roun trip time (ms) to sen one ata packet an receive an acknowlegment packet in return. The DTO-ACK moel formulates (3-7) in orer to achieve optimum fairness in the process of sening ata packets an receiving an acknowlegement packet from each respective source noes in a certain network. DACKn = Fair cycle time (t) - RTTn (3) Where DACKn is the TCP elaye acknowlegement timeout for n number of noes in a eicate path (O/Even) base on

3 Draft Version - Brunel University, Lonon the hop count, Where Fair cycle time (t) is from (2) an RTTn is the roun trip time base on the hop count to the estination noe. RPPNMax = SimDuration / DACK Min (4) Where RPPNMax is the achievable maximum number of receive packets per noe in the network, SimDuration is the simulation uration in secons an DACK Min is the minimum elaye acknowlegement timeout calculate from the formulation in (3). RPPNMin = SimDuration / DACK Max (5) Where RPPNMin is the achievable minimum number of receive packets per noe in a certain network, SimDuration is the simulation uration in secons an DACK Max is the maximum elaye acknowlegement timeout calculate from the formulation in (3). TMax = (RPPNMax 8 Pkt size) / SimDuration (6) Where TMax is the achievable maximum throughput per noe in the network base on (4). This throughput is achievable for a respective noe to transmit successful packets to a estination noe when the chronology of transmission an the elay acknowlegement timeout is consistent for the entire SimDuration. TFair = (RPPNMin 8 Pkt size) / SimDuration (7) Where TFair is the achievable fair throughput per noe in the network base on (5). The TFair is base on the lowest achievable throughput in the network that best escribes the throughput capacity of the furthest noe in the network from the estination chronology of transmission an the elay acknowlegement timeout is consistent for the entire SimDuration. The value obtaine from (7) was use in limiting the total packets generate by each eicate source noes in the network to achieve the esirable throughput fairness. This formulation also improves the instability when there are packets roppe uring the simulation uration by narrowing the throughput maximum an minimum gap. The variation on TMax an TFair is significant on a network with a large number of noes (source). The DTO-ACK formulation (1-5) will result in throughput fairness for all source noes in the wireless network. The RTT value use in all the above formulation is 14.1 ms. The RTT value is an average value of ten runs obtaine through a simple simulation setup of two noes istribute in a istance of 100 meters using DI-LSRP. With a lower value of RTT woul results in a higher throughput value. The throughput fairness ratio of (1:1) can be achieve even with increasing number of source noes in a certain wireless network by using the formulation shown in (1-5). In a general TCP scenario, the RTT is efine as one cycle time from On/En to sen a ata packet to ND an receive an acknowlegement packet. Thus, both ata an acknowlegement packet take place in one RTT value where else for DTO-ACK it works in a series of ata packets an then followe by a series of acknowlegement packets as shown in Fig. 2. A simple five noes with four source noes (O path: O1 an On, Even path E1 an En) an one estination noe as ND with DTO-ACK is shown in Fig. 2. DTO-ACK moel can achieve optimum fairness among all source noes as shown in Fig. 2 by planne timely sequence of ata packets with pre-calculate elaye acknowlegement packets between a source an a estination noe. noe. This throughput is achievable for a respective noe to transmit successful packets to the estination noe when the 1 Sum H cycle time Data packets En On E1 O1 Sequence no. 1 Sequence no. 2 Sequence no. 3 Sequence no. 4 Acknowlegment packets Ack-En Data-E1 Ack-En Ack-E1 Data-E1 Data-O1 Ack-O1 Data-O1 Ack-O1 Ack-E1 ND 1st cycle Data-O1 1 hop Data-E1 2 hops 3 hops 4 hops Ack-En 5 hops 2n cycle Fig. 2: Two (1 Sum H cycle) timing iagram DTO-ACK moel with On/En number of source noes an ND as the estination noe.

4 Draft Version - Brunel University, Lonon All source noes (O1/E1/On/En) will initiate the first ata packet (Data-O1/Data-E1//) transmission to the estination noe (ND) at the start of each 1 Sum H cycle (t). The ata packets (Data-O1/Data-E1) from source noes (O1/E1) requires one hop to the esire estination noe (ND) at uration of one RTT where else for the ata packets (/) from source noes (On/En) requires two hops through the intermeiate noe (O1/E1) to reach the estination noe (ND) at with two RTT value. Base on the DTO-ACK moel, the first sequence of acknowlegement packets (/Ack-En) will be generate at the estination noe (ND) with two hops require through the intermeiate noe (O1/E1) to reach the source noes (On/En) at the ege of six RTT value. In the secon sequence of acknowlegement packets (Ack-O1/Ack-E1) from the estination noe (ND) requires only one hop to reach the source noes (O1/E1) at the ege of six RTT value. From the ual path sequence (O an Even route) shown in Fig. 1, this woul take six RTT (3RTT + 3RTT) value for 1 Sum H cycle (t) to achieve the equal ata transmission opportunity to all source noes (O1/E1/On/En) in the network to sen a ata packet an receive an acknowlegement packet from the estination noe (ND). Using DTO-ACK moel in TCP agent will ensure throughput fairness among all source noes in a linear topology environment which will suit well in an oil an gas pipeline monitoring application. The key parameter to be efine is the total number of hops to a estination noe in a ual path with an appropriate RTT value which is base on istance will benefit on the network equality. Changes in the scenario where a scalable WSN with selective source noes is applicable with DTO-ACK moel by assigning the right Sum H in (1) in orer to overcome with banwith reservation for passive noes (non-source noes). simulation are stationarily locate with a fixe uniform istance as shown in Table I for the entire simulation uration with only one estination noe. TABLE I NS2 SIMULATION PARAMETERS FOR DTO-ACK AND COMPARED MODEL Parameters Channel type Raio propagation moel Value Wireless channel Two ray groun MAC type Interface queue type Drop Tail / PriQueue Source noes 19, 39, 59, 79, 99 Max packet in ifq Agent type Traffic type RX Thresh/ CS Thresh Packet size 50 (packets) DTO-ACK /100 (packets) Others Transmission control protocol (TCP) - Reno Constant bit rate (CBR) 125 meters/250 meters 512 bytes Dual Interleaving Linear Static Routing Protocol (DI-LSRP) is a preefine ual path routing protocol with reuce control packets. The simulation was teste with AODV, DSDV, FRP using the metho propose in [7] an DI-LSRP using DTO-ACK are evaluate on the following metrics: A. Delivery ratio: A percentage of receiving packets over sen packets in a network is an essential performance inicator for the reliability of a certain wireless network. With the implementation of DTO-ACK in a linear wireless network, helps to retain the elivery ratio at a constant rate of always above 99 % from a small network size of 24 source noes to a large network size of 120 source noes as shown in Fig.3. IV. SIMULATION ANALYSIS AND DISCUSSION This section of the paper illustrates results on overall network performance with optimum throughput fairness with the propose DTO-ACK moel. All the presente results are obtaine through simulation using Network Simulator 2 (Version 2.35) [18]. The DTO-ACK moel was incorporate with DI-LSRP was compare with reactive routing algorithm AODV, proactive routing algorithm DSDV an FRP using the metho propose in [7]. The basic simulation setting in Network Simulator 2 is as tabulate in Table I. All results are from an average value of five multiple runs with see values (1-15) over a simulation uration of 500 secons. Two ranom functions (uniform) were use to re-create the reallife application of the teste metho, where the first ata packet start time for each source noe is generate using a custom ranom function with a range between secons. Another ranom function was use to ranomly select a non-sequence source noe chronology for ata transmission. Noes in the Delivery ratio (%) AODV DSDV FRP DI-LSRP Fig. 3: Graph of elivery ratio (%) over number of source noes The propose metho reuces the number of roppe packets which is a esirable parameter in any wireless sensor network.

5 Draft Version - Brunel University, Lonon B. Fairness inex: In a linear topology with n number of source noes with only one estination noe has a crucial stability factor ue to unfairness within the wireless network. The scalar measurement of resources (throughput) allocation iscrimination among all source noes [7, 16]. The Jain s fairness inex [19, 20] is as escribe in (8) is use in calculating the factor of fairness in a multi-hop network. The highest esirable throughput fairness inex of 1.0 represents that a network has achieve 100% normalise throughput fairness inex. = (8) Where ni is the normalise throughput for n number of flows an for N is the number of noes in the network. In a small size linear network with a low ensity of sensor, noe fairness is harly visible since the effects as mentione in Section II of this paper. Throughput fairness inex in a network with DTO-ACK outperforms all the other routing protocol from a small network size of 24 source noes to a large network size of 120 source noes as shown in Fig.4. AODV Throughput fairness inex 1.0 DSDV FRP DI-LSRP the increasing number of source noes in the teste simulation environment. The greatest challenge in a linear wireless sensor network is in achieving optimum throughput fairness inex among all source noes in a certain network. One of a measurement inicator for throughput fairness inex is throughput, where equality an performance plays a major role in any WSN. The next set of results will escribe the performance metric in all the simulate environment to inicate a viable solution for fairness as well as performance. C. Throughput: The average throughput overall flows in the network can be calculate as given in [7, 21]. When network performance is a concern in a WSN, a network with the ability to achieve higher throughput are always a esirable goal. Referring to Fig. 5, throughput (Kbps) in a network with DTO-ACK outperforms all the other routing protocol from a small network size of 24 source noes to a large network size of 120 source noes. Since the formulation in DTO-ACK is base on ual interleaving route which reuces the ata travel cost (total hop metric) hence, increases the ata transfer rate between source noes to the estination noe by two fols in the teste simulation environment. The DTO-ACK moel enables high ata transfer rate with the network resources by retaining optimum throughput fairness inex in a linear WSN. AODV 30 DSDV FRP DI-LSRP 25 Throughput (Kbps) For the other compare routing protocol, only FRP with the moel in [7] retains the elivery ratio at a constant rate of above 99 % equal to the propose metho where else for AODV an DSDV the elivery ratio ecrease with the increasing number of source noes. Utilising a TCP elaye acknowlegement mechanism woul a a great value towars successful ata packet elivery in a linear wireless network Fig. 5: Graph on throughput (Kbps) over number of source noes Fig. 4: Graph of throughput fairness inex over number of source noes For the other compare routing protocol, only FRP with the moel in [7] achieve higher throughput fairness inex similar to DI-LSRP, but when compare in terms of ata rate this is relatively low. The higher throughput value as shown in Fig. 5 has a corresponing effect towars the variation in throughput fairness inex between DI-LSRP an FRP. Where else, the throughput fairness inex for AODV an DSDV ecrease with D. En to en elay: En to en elay is the average total time taken to transmit ata over all the flows in the network [21-23]. Referring to Fig. 6, the en to en elay with DTO-ACK moel ha relatively outperforme the other routing protocol from a small network size of 24 source noes to a large network size of 120 source noes. The value of fairness inex from Fig. 4 an throughput from Fig. 5 has a corresponing effect on en to en elay with DTO-ACK moel. Even though DSDV has a lower en to en elay, but this has a corresponing effect in poor fairness inex as in Fig. 4 an low throughput as in Fig. 5 hence, DI-LSRP has comparatively lower en to en elay when compare to the other relate metrics. The increase in the en to en elay is proportional to the increasing receive ata rate as

6 Draft Version - Brunel University, Lonon well as the increasing istance between source noes an the estination noe. The other routing protocol has relatively higher en to en elay with lower ata rate. Every routing protocol has its rawback towars en to en elay which mainly relates to the number of control packets to support network connectivity. With DTO-ACK moel, en to en elay is fairly low even with the higher RTT value compare in [7]. En to en elay (ms) Fig. 6: Graph on en to en elay (ms) over number of source noes E. Routing overhea: In a WSN, normalise routing overhea is efine as the total number of routing packets use in the network over the total receive ata packets for the entire simulation uration [23] which can be calculate by (9). = AODV DSDV FRP DI-LSRP Where NRO is normalise routing overhea is another crucial factor in any wireless sensor network as there is a queue length for biirectional packets an performance factor relate to this metric. Referring to Fig. 7, normalise routing overhea with DTO-ACK moel is fairly low compare to the other routing protocol from a small network size of 24 source noes to a large network size of 120 source noes. The normalise routing overhea increases relatively with the increasing number of source noes in a network. Each routing protocol has a unique characteristic which contributes towars the increasing number of routing overhea in a network. The implementation of DTO-ACK an the metho propose in [7] influences the increasing number of routing overhea even in an ile or waiting between ata transmission state. Hence, a routing protocol with low control packets will reuce the effect on routing overhea such as in DI- LSRP. In general, by controlling the normalise routing overhea woul result in better throughput fairness inex as in Fig. 4 an overall network performance as in Fig. 5 to Fig.6 in a WSN particularly in a long-range wireless network. (9) Routing overhea (packets) AODV DSDV FRP DI-LSRP Fig. 7: Graph on normalise routing overhea (Joules) over number of source noes F. Passive noes: Passive noes are consiere noes with zero ata transfer to the estination noe in a network. Passive noes create communication breakown from a certain point in a network ue to a single point failure factor in a linear topology. A certain network with a higher number of passive noes will be a waste of available resources, especially when working in a network of high traffic an limite energy source. With the implementation of DTO-ACK an the metho propose in [7] eliminates the problem with passive noes in the teste environment. V. CONCLUSION This paper has highlighte a research conucte to achieve better throughput fairness inex an overall network performance on a linear one-tier topology using DTO-ACK in TCP agent. The istance between source an a estination noe in a linear topology is a significant factor towars imbalance in network resources which results in complete ata flow starvation. Data flow starvation leas towars performance relate issues in WSN, particularly in a linear topology. The mathematical formulations in DTO-ACK moel for setting an appropriate elaye acknowlegement timeout for respective source noes with low avertise winow size woul retain the throughput fairness inex at an optimum level. DI-LSRP using the propose DTO- ACK moel was verifie through simulation where significant improvements in throughput fairness inex an overall network performance were achieve when applie in TCP agent. DTO- ACK has a unique formulation to retain throughput fairness along with moerate throughput an low en to en elay for a variable linear network size which makes it a viable solution for a pipeline network. ACKNOWLEDGMENT The authors woul like to thank Ministry of Eucation Malaysia an University Teknikal Malaysia Melaka for their

7 Draft Version - Brunel University, Lonon financial support an assistance for the research conucte in Brunel University Lonon. REFERENCES [1] A.C. Azubogu, V.E. Iigo, S.U. Nnebe, O.S. Oguejiofor an E. Simon, "Wireless Sensor Networks for Long Distance Pipeline Monitoring," in Proceeings of Worl Acaemy of Science, Engineering an Technology, pp. 86, [2] S. Savazzi, S. Guariano an U. Spagnolini, "Wireless sensor network moeling an eployment challenges in oil an gas refinery plants," International Journal of Distribute Sensor Networks, vol. 2013, [3] H. Devol, Oil an gas prouction hanbook: an introuction to oil an gas prouction, Lulu. com, 2013,. [4] A.M. Al-Jubari an M. Othman, "A new elaye ACK strategy for TCP in multi-hop wireless networks," in Information Technology (ITSim), 2010 International Symposium in, pp , [5] L. Venkatesan, S. Shanmugavel an C. Subramaniam, "A survey on moeling an enhancing reliability of wireless sensor network," Wireless Sensor Network, vol. 5, pp. 41, [6] Z. Wang, X. Zha an X. Qian, "The application an issuse of linear wireless sensor networks," in System Science, Engineering Design an Manufacturing Informatization (ICSEM), 2011 International Conference on, pp. 9-12, [7] T. Hou, C. Hsu an C. Wu, "A elay base transport layer mechanism for fair TCP throughput over multihop wireless mesh networks," International Journal of Communication Systems, vol. 24, pp , [8] H. Shi, R.V. Prasa, E. Onur an I.G. Niemegeers, "Fairness in wireless networks: Issues, measures an challenges," Communications Surveys & Tutorials, IEEE, vol. 16, pp. 5-24, [9] W.R. Stevens, "TCP slow start, congestion avoiance, fast retransmit, an fast recovery algorithms," [10] R. Oura an S. Yamaguchi, "Fairness comparisons among moern TCP implementations," in Avance Information Networking an Applications Workshops (WAINA), th International Conference on, pp , [11] E. Blanton, M. Allman an V. Paxson, "TCP Congestion Control," [12] V. Paxson, M. Allman, J. Chu an M. Sargent, Computing TCP s Retransmission Timer, [13] B. Chen, I. Marsic, H. Shao an R. Miller, "Improve elaye ack for tcp over multi-hop wireless networks," in Wireless Communications an Networking Conference, WCNC IEEE, pp. 1-5, [14] P.T. Giang an P.M. Vi, "Cross layer esign to enhance TCP performance in multi-hop a hoc networks," in Avance Technologies for Communications (ATC), 2013 International Conference on, pp , [15] B. Wang, Y. Chen, Y. Hu an W. Feng, "TCP Vegas-W: Enhance TCP Vegas congestion avoiance mechanism [J]," Journal of Computer Applications, vol. 9, pp. 058, [16] A.M. Al-Jubari, M. Othman, B.M. Ali an Hami, Nor Asilah Wati Abul, "An aaptive elaye acknowlegment strategy to improve TCP performance in multi-hop wireless networks," Wireless Personal Communications, vol. 69, pp , [17] A.M. Al-Jubari, M. Othman, B.M. Ali an Hami, Nor Asilah Wati Abul, "TCP performance in multi-hop wireless a hoc networks: challenges an solution," EURASIP Journal on Wireless Communications an Networking, vol. 2011, pp. 1-25, [18] T. Issariyakul an E. Hossain, Introuction to network simulator NS2, Springer Science & Business Meia, 2011,. [19] R. Jain, D. Chiu an W.R. Hawe, A quantitative measure of fairness an iscrimination for resource allocation in share computer system, Eastern Research Laboratory, Digital Equipment Corporation Huson, MA, [20] A.B. Seiq, R.H. Gohary, R. Schoenen an H. Yanikomeroglu, "Optimal traeoff between sum-rate efficiency an Jain's fairness Inex in resource allocation," IEEE Transactions on Wireless Communications, vol. 12, pp , [21] M. Cao, L.T. Yang, X. Chen an N. Xiong, "Noe placement of linear wireless multimeia sensor networks for maximum network lifetime," in Avances in Gri an Pervasive Computing, Springer, 2008, pp [22] S. Tyagi an R. Chauhan, "Performance analysis of proactive an reactive routing protocols for a hoc networks," International Journal of Computer Applications, vol. 1, pp , [23] Y. Khamayseh, G. Obieat an M.B. Yassin, "Mobility an Loa aware Routing protocol for a hoc networks," Journal of King Sau University- Computer an Information Sciences, vol. 23, pp , 2011.