# Introduction obile Ad-hoc Network(MANET) is an autonomous infrastructure-less system with wireless links connecting mobile nodes each operating as an end system as well as a router to forward packets and are free to move and form a network without any base stations. As mobile nodes communication is faster, it has a wide application in military applications, emergent operations, personal electronic device networking, and civilian applications like an ad-hoc meeting or an ad-hoc classroom. MANET's special characteristics like dynamic topologies, bandwidth constrained, variable capacity links, energy constrained operation, limited physical security make it susceptible to various attacks. In addition, jamming occurs in unreliable wireless links and eavesdropping due to their inherent broa dcast nature. Mobile devices' bandwidth, computing power, as well as battery power constraints causes application-specific trade-offs between devices' security and resource consumption. Behavior anomalies like advertising bogus routes are difficult to detect due to mobility [1]. The Transmission Control Protocol (TCP), the most commonly used reliable transport protocol in the internet, provide end-to-end reliable transmission as well as fair congestion control so as to share network resources efficiently [2]. Congestion control is a necessity for MANET since it operates in bandwidth constrained conditions. TCP, in today's internet has poor performance and in MANETs is worsen by contention with increasing UDP-based high priority multimedia traffic and the class differentiation introduced in current QoS protocols, which results into TCP starvation and increased spurious timeouts [3]. The TCP usage over wireless networks rise to problems due to the different characteristics of wireless links with respect to wired ones, in terms of less reliability and time-variant behavior, fading / shadowing problems, node mobility, hand-offs, limited available bandwidth and large RTTs [5]. TCP performs reliably in traditional wired and stationary networks, but network congestion induces losses. Moreover, if the router waiting queues are full or nearly full, the packets received are dropped, thereby wired network losses are viewed as an indication of congestion. But in wireless networks the losses occur frequently for various causes such as interference, node mobility or poor link quality. In MANETs, data transmission failures due to node mobility, interference, poor link quality, etc lead often to the congestion control activation by TCP protocol unnecessarily which degrades the performance of TCP in MANETs [2,4,6] . The TCP reliability guarantees, TCP sender will forward each packet issued by it to the addressee node by a system of acknowledgements [4]. In TCP protocol, cumulative acknowledgements (ACKs) are sent by the TCP receiver for successfully received segments so as to let the TCP sender to find out the segments which have successfully received. The TCP sender determines the loss of packet either by several duplicate ACKs arrival, triggering a fast retransmission, or by the absence of an ACK for a timeout interval, thereby the TCP retransmits the lost packets, and simultaneously invokes congestion control by reducing its congestion window size and backing off its retransmission time for compensating packet losses. These reduce the level of congestion on the intermediate links [7]. The TCP Acknowledgements may return collide with their corresponding data packets in particular if the number of network nodes becomes vital. Such a large number of TCP ACK packets come into contention with TCP packets leads to intra-flow contention [4]. As an extension to this work, we propose to design a smart acknowledgement distributed channel access scheme for TCP in order to reduce the delay and overhead of sending TCP acknowledgements. # II. # Literature Review Chien-Chia Chen et al [2] presented Combo Coding, a novel coding scheme combining intra-and inter-flow coding and a novel loss adaptation algorithm was featured. Combo Coding decreases the data and ACKs' interference within a TCP session taking advantage of the benefits of both types of codes, and also robustness to high link losses is exhibited. 2 Mbps good put is achieved by Combo Coding successfully with 30% per link packet loss rate; whereas TCP-New Reno with no coding delivers only 200 Kbps. Combo Coding over 2 paths provides a lower good put than that over a single path due to a higher overhead due to extra contentions as in the 20s-50s interval. Sofiane Hamrioui et al [4] suggested an improvement of the Transmission Control Protocol (TCP), called Improvement of the Acknowledgement mechanism of TCP (IA-TCP), for MANET's better performance. Based on number of nodes, mobility and the communication distance between these nodes, IA-TCP delays TCP's acknowledgements packets. However, only throughput and end-to-end delay parameters are concerned. However, there is degradation of the TCP parameters performance. Mohammad Amin Kheirandish Fard et al [10] proposed an end-to-end sender-side approach classifying congestion loss, wireless channel loss and link failure loss by queue usage estimation. Queue usage is calculated using relative One-way Trip Time. Congestion loss is estimated using packet losses with queue usage rate greater than predefined threshold whereas non-congestion loss is when queue usage is less than threshold. Non congestion loss recognized by three duplicate ack labels loss caused by wireless channel error as it indicate the route existence between communicating end point that duplicate ACKs moved along. Deguang Le et al [11] proposed a new crosslayer approach introducing a mobility detection element in the network layer to communicate with the transport layer for optimizing TCP operations. This approach preserves the end to end TCP semantics since end point changes are only made. Unlike other networks utilizing either transport or network layer alone without much cross-layer cooperation, this approach enable mobility information in TCP. However, RTO is reduced significantly compared to the standard TCP/MIPv6 as soon as the handover termination. Consolee Mbarushimana et al [12] proposed a novel TCP-friendly scheme, IEDCA, for enhancing IEEE 802.11e EDCA mechanism by assigning the highest priority to TCP acknowledgement packets. The proposed scheme improves TCP performance significantly while having very negligible effects on voice traffic. However, the voice traffic dropped by buffer overflow or when the retry threshold is exceeded is higher in EDCA. Toktam Mahmoodi et al [13] presented and assessed a cross-layer solution for a node for faster adaption of lower-layer characteristics like the coding rate and local ARQ retransmissions threshold according to the detected TCP flavor, so as to optimize the end-toend performance of the download for that utilized flavor. The proposed scheme has considerable potential to improve the overall download throughput, without burdening the server and requiring no changes to existing TCP implementations. However, end-to-end throughput decreases with increasing propagation distance. Myungjin Lee et al [14] proposed a path recovery notification (TCP-PRN) mechanism for avoiding performance degradation while handoff. Despite of being attained performance improvement while handoff, Freeze-TCP deployment in real environment is difficult due to obstacles in detecting accurate handoff time and the vulnerability of high variation in the round trip time (RTT). The proposed protocol restore congestion window avoiding it to reduce or immediately initiate slow start algorithm to recover lost packets. However, the handoff prediction is difficult and the variation of RTT is higher. Ammar Mohammed AI-Jubari and Mohamed Othman [15] proposed a new delay-ACK algorithm for TCP performance enhancement over multi-hop wireless networks. According to factors like TCP startup phase, transmission rate, packet loss event, and number of hops, the optimal delayed ACK window is achieved dynamically as like before. The dynamic adaptive strategy's objective is enabling TCP receiver to adjust itself in terms of the data to ACK ratio. The scheme enhances TCP performance, attaining up to 233 % performance gain, over multi-hop wireless networks, compared to the regular TCP. However, the burst transmission occurs at the sender triggered by delayed In our previous work [18], a cross-layer based approach for improving TCP performance in Multihop Mobile Ad-hoc Networks (MANETs) is proposed. The proposed mechanism triggers congestion whenever the channel occupied ratio (COR) reaches a maximum threshold value and the received signal strength is less than a minimum threshold value. Following it, congestion control scheme controls the data sending rate of the sender by determining available bandwidth, delay of its link and COR. Further, a fair resource allocation scheme is put forwarded. Volume XV Issue VII Version I Year 2015 ( E ) ACK. The burstiness increases the packet loss and potentially affects TCP performance. # III. # Proposed Solution a) Overview The delay in packet transmission leads to loss of packet as well as throughput degradation. Hence we use a separate access category for data less TCP acknowledgement packets and assumed higher priority to them [12]. Hence delay in packet transmission can be reduced and packets are acknowledged instantaneously. Here higher congestion window value is attained than by TCP over Enhanced Distributed Channel Access (EDCA). Further improvements can be made by adjusting the delay window size by considering channel conditions like transmission rate, slow start phase, number of hops and packet lost event [15]. Here the number of acknowledgement packets can be reduced by generating the acknowledgement packets after attaining optimal dynamic delay window. Therefore receiver can adapt according to various delays. # b) Estimation of Metrics i. Probability of Successful Transmission : Most of the previous studies propose the probability of successful transmission based on analytical skills to compute delay and throughput that can be attained by different traffic categories in IEEE 802.11e for which they mainly considered Markov model. In this type of model successful transmission is given as below: ?= 2 1+V+pV ? (2p) i n ?1 i=0(1) V represents initial contention window during backoff stage, p represents the conditional collision probability, and n represents maximum backoff stage. Also the transmission probability k ? of a station of Access Category (AC) k can be given as below: ? k = 2 1+V k +p k V k ? (2p k ) i n k ?1 i=0(2) Based on above equation, similar expression can be obtained for throughput T k which is attainable by an access category k as below T k = n k ? k (l?p k )l p(t)H (t) +p(c)H (s) +p(e) ?(3) Where l represents average transmission time of payload for access category 'k'. p(t), p(c) and p(e) are the probabilities of a slot time contains successful transmission, collision, and channel being idle respectively. H(s), H(t) and ? represents average slot length of successful transmission, collision and channel being inactive respectively. From the above expressions, it can be observed that probability of successful transmission for any AC is inversely proportional to contention window V. Hence station with traffic in high priority AC will have low values of V which in turn will have higher chance of transmitting their data than stations with best effort and will have better throughput. But this case is not applicable for TCP traffic. For TCP traffic, a packet is considered to be successfully transmitted in case TCP ACK is received by the sender before the retransmission time is over. Based on the probability idea of successful transmission given by equation ( 2), successful transmission probability in case of TCP traffic is a combination of probability of successfully transmitted data packet as well as probability of successfully received ACK packet. Hence, for TCP traffic belonging to AC1, probability of successful transmission can be given as below: ?? TCP = 4 ?1+V 1 +p 1 V 1 ? (2p i ) i n 1 ?1 i=0 ? 2(4) Here assume that data and ACK have the same transmission probability, since they belong to the same access category and moreover delayed ACK option is not used. Hence TCP throughput is different from any other best attempt throughput which is accessible by equation ( 3), as it depends on the value ?TCP which is given in the equation ( 4) ii. Channel Occupied Ratio (COR) COR is defined as the ratio of total lengths of busy periods to the total transmission time. . Consider T T denotes the total transmission time and T b represents total length of busy periods. Hence COR can be given as below: ?????? = ?? ?? ?? ?? (5) By considering channel utilization factor, a threshold value Th COR , it can be given as below: ?????? ? C U (COR ? Th COR ) (6) Where, C U denotes channel utilization factor and it is the measure of ratio of channel busyness time for successful transmissions to the total time T T . # c) Smart Acknowledgement Distributed Channel Access Scheme This section describes the proposed Smart Acknowledgement Distributed Channel Access (SADCA). The main aim of the proposed technique is to maximize the throughput without creating any delay in the network. In this technique, an additional fifth access category saved for data less TCP ACK is defined. This new AC is assigned with highest priority. By placing data less TCP ACKs in top priority access category, probability of transmitting them on time is also maximized. Since the TCP ACK transmission probability tends to 1, transmission probability of TCP traffic can be expressed by equation (2). In similar way, the possible TCP throughput can be obtained by equation( 3) with the exception that H s increases by the value equal to time required to transfer both TCP ACK as well as MAC control packets that go along with it. By utilizing this method, the probability that all the TCP packets transmitted get acknowledged are increased. The underlying principle of this method make use of transport layer to control the volume of TCP ACKs on the channel. The volume of TCP ACKs permitted on the channel matches the volume of TCP data which is successfully transmitted and it is independent of the type of other traffic struggling for the same medium. To avoid any kind of congestion due to retransmission COR is validated inside the network using equation ( 5) and ( 6), if in case the delay due to congestion is found. # The implementation of the SADCA can be explained in steps as below: Step 1 : In the first step, type of service (ToS) is usually applied at the application layer, however TCP ACKs originate from the transport layer, hence transport layer is responsible for assigning a proper ToS to them. Step 2 : Since transport layer automatically assign the best effort ToS to TCP ACKs, a different ToS need to be assigned to them to distinguish them from best effort traffic. Step 3 : To make the process simple, ToS value ( 224) is assigned to the data less TCP ACK at the transport layer when they are about to sent to the lower layer. Step 4 : The MAC layer describes 8 user priority levels which are used to arrange the different data packets into different ACs. Step 5 : By considering 224 as their transport layer ToS, TCP ACK will be allocated to the user 7. Step 6 : Also a fifth AC is defined corresponding to MAC user priority value of 7 and hence collect the TCP ACKs. The mapping of client priority (CP) to AC in SADCA is shown in Table 1. Hence an additional traffic queue is obtained in SADCA as shown in Fig. 1. The proposed aim is to allow TCP ACKs priority to the medium. After that priority is assigned to the new AC having smaller values of AIFS, CWmin and CWmax compared to the existing higher priority voice traffic as shown in Table 2. The improvement in the proposed technique can be shown with the following example: In case, TCP ACK is produced at the same time as the voice traffic, then station T S gains the medium contention because these TCP ACKs are in an AC with higher priority than voice. Hence it can be said that T S have more chance to transmit the TCP ACK before timeout and hence avoid the retransmission at the source. If the TCP ACK is generated before timeout, and then it avoids any kind of retransmission at the source. Also if TCP ACK is generated after the VoIP conversation started, then it will have a good opportunity to compete for the medium at the end of the next TXOP of the VoIP stations. Hence it can be said that, there is a high chance that ACK will be received within the RTO limit. Also retransmission can be avoided. Fig. 4 shows the delivery ratio of SADCA-TCP and TCP techniques for different number of flows scenario. We can conclude that the delivery ratio of SADCA -TCP has 3% of higher than TCP. Fig. 5 shows the throughput of SADCA-TCP and TCP techniques for different number of flows scenario. We can conclude that the throughput of SADCA-TCP has 45% of higher than TCP. The simulation settings and parameters are summarized in Table 4. Fig. 8 shows the delivery ratio of SADCA-TCP and TCP techniques for different number of flows scenario. We can conclude that the delivery ratio of SADCA -TCP has 2% of higher than TCP. Fig. 9 shows the throughput of SADCA-TCP and TCP techniques for different number of flows scenario. # IV. # Simulation Results We can conclude that the throughput of SADCA-TCP has 37% of higher than TCP. V. # Conclusion In this paper we have proposed a Smart Acknowledgement Distributed Channel Access (SADCA) scheme for TCP in MANETs. In the proposed scheme, first a separate access category for data less TCP acknowledgement packets is used and also it is assigned with highest priority. As the TCP ACK is assigned with new AC with highest priority, it has maximum probability of successful transmission by making efficient utilization of contention window during backoff stage. In this way, delay during transmission of packet can be reduced and also packet can be acknowledged immediately. Also, to increase the performance, delay window size can be adjusted for optimization purpose by considering the parameters such as transmission rate, number of hops, and congestion occupied ratio (COR). Hence the proposed scheme helps to avoid any kind of delay and overhead for sending TCP acknowledgement 12![Fig. 1: Node within Transmission Range with Multiple priority queues in SADCA](image-2.png "Fig. 1 :Fig . 2 :") 34![Fig. 3: Flows Vs Delay](image-3.png "Fig. 3 :Fig. 4 :") 5![Fig.5: Flows Vs Throughput Fig.3shows the delay of SADCA-TCP and TCP techniques for different number of flows scenario. We can conclude that the delay of SADCA-TCP has 35% of less than TCP.Fig.4shows the delivery ratio of SADCA-TCP and TCP techniques for different number of flows scenario. We can conclude that the delivery ratio of SADCA -TCP has 3% of higher than TCP.](image-4.png "Fig. 5 :") 6![Fig. 6 : Dynamic Random Topology In the dynamic random topology, 50 mobile nodes are deployed randomly in a 1500 meter x 300 meter region as shown in Fig. 6. All nodes have the same transmission range of 250 meters. The random way point model of ns-2 is used as the mobility model in which the pause time of the node is 5 seconds and speed is 10 m/s. There are 10 traffic flows between different set of source and destination pairs.The simulation settings and parameters are summarized in Table4.](image-5.png "Fig. 6 :") 789![Fig. 7 : Flows Vs Delay](image-6.png "Fig. 7 :Fig. 8 :Fig. 9 :") 1Client Priority(CP)Access Category(AC)Designation10Background20Background01Best Effort31Best Effort42Video52Video63Voice74TCP ACK81COR 2in SADCAACCW mi CW max AIFSNTXOP limitVoice71523264Video153126016Best Effort31102330Backgroun31102373264d TCP ACK3713264 Smart Acknowledgement Distributed Channel Access Scheme for TCP in Manetsa) Simulation Setup NS-2 simulator is used to simulate the Smart Acknowledgement Distributed Channel Access scheme for TCP (SADCA-TCP) protocol. In our simulation, the channel capacity of all mobile hosts is Queue 0 Backoff AIFS AC0 Scheduler (Resolve virtual Collision) Transmission Queue 1 Queue 2 Queue 3 Backoff AIFS AC2 Backoff AIFS AC3 Backoff AIFS AC4 Backoff AIFS AC1 set to 2 Queue 4 Background Best Effort Video Voice TCP ACK 3No. of Nodes9Area Size1700 X 300Mac8021.11Radio Range250mSimulation Time300 secTraffic SourceFTPPacket Size1000 bytesNo. of Flows1 to 8Routing ProtocolAODVThe simulation results are given below: 4No. of Nodes50Area Size1500 X 300Mac8021.11Radio Range250mSimulation Time300 secPause time5 secondsSpeed10 m/sTraffic SourceFTPPacket Size512 bytesNo. of Flows1 to 10Routing ProtocolAODVThe simulation results are given below: © 2015 Global Journals Inc. (US) 1 * A trust model based on fuzzy recommendation for mobile ad-hoc networks JunhaiLuo XueLiu MingyuFan 2009 * Combo Coding: Combined intra-/inter-flow network coding for TCP over disruptive MANETs Chien-ChiaChen CliffordChen SoonYOh Joon-SangPark MarioGerla MYSanadidi Journal of Advanced Research 2011 * E-TCP: Enhanced TCP for IEEE802.11e Mobile Ad Hoc Networks ConsoléeMbarushimana AliShahrabi 15th International Conference on Parallel and Distributed Systems 2009 * IA-TCP: Improving Acknowledgement Mechanism of TCP for better performance in MANET SofianeHamrioui MustaphaLalam PascalLorenz International Journal on New Computer Architectures and Their Applications (IJNCAA) 2012 * A Crosslayer Scheme for TCP Performance Improvement in Wireless LANs DzmitryKliazovich FabrizioGranelli IEEE Global Telecommunications Conference, GLOBECOM'04 2004 2 * A Cross layer solution to improve TCP performances in Ad hoc Wireless Networks YassineDouga MalikaBourenane IEEE International Conference on Smart Communications in Network Technologies (SaCoNeT) 2013 1 * Delay Performance of the New Explicit Loss Notification TCP Technique for Wireless Networks WenqingDing AbbasJamalipour IEEE Global Telecommunications Conference, GLOBECOM'01 2001 6 * A Smart TCP Acknowledgement Approach for Multihop Wireless Networks TorstenRuy De Oliveira Braun IEEE Transactions On Mobile Computing 6 2 February 2007 * Improving Transport Layer Performance in Multihop Ad Hoc Networks by Exploiting MAC Layer Information HongqiangZhai XiangChen YuguangFang IEEE Transactions On Wireless Communications 6 5 May 2007 * Packet Loss Differentiation of TCP over Mobile Ad Hoc Network Using Queue Usage Estimation MohammadAmin Kheirandish Fard SasanKaramizadeh MohammadAflaki IEEE 3rd International Conference on Communication Software and Networks (ICCSN) Xi'an 2011 * A Cross-Layer Approach for Improving TCP Performance in Mobile Environments DeguangLe XiaomingFu DieterHogrefe Wireless Personal Communications Feb 2010 52 * A Cross-Layer Support for TCP Enhancement in QoS-Aware Mobile Ad Hoc Networks ConsoleeMbarushimana AliShahrabi TomBuggy proceeding of the 11th International Symposium on Modeling Analysis and Simulation of Wireless and Mobile Systems eeding of the 11th International Symposium on Modeling Analysis and Simulation of Wireless and Mobile SystemsMSWiM, Vancouver, British Columbia, Canada October 27-31, 2008 * Cross-Layer Optimization of the Link-Layer based on the Detected TCP Flavor ToktamMahmoodi OliverHolland VasilisFriderikos HamidAghvami IEEE 19th International Symposium on Personal, Indoor and Mobile Radio Communications PIMRC, Cannes 2008 * A cross-layer approach for TCP optimization over wireless and mobile networks MyungjinLee MoonsooKang MyungchulKim JeonghoonMo Computer Communications 2008 * A New Delayed ACK Strategy for TCP in Multi-hop Wireless Networks AmmarMohammed Ai-Jubari MohamedOthman Information Technology International Symposium (ITSim) Kuala Lumpur 2010 2 * TCP with Delayed Ack for Wireless Networks JiweiChen YengZhong Lee MarioGerla MYSanadidi 2006 Networks and Systems BROADNETS 3rd Broadband Communications * Reducing the TCP Acknowledgement Frequency SaraLandström Lars-ÅkeLarzon ACM SIGCOMM Computer Communication Review * A cross-layer approach for congestion DSunitha ANagaraju GNarsimha Multi hop Mobile Ad hoc Networks" Computing for Sustainable Global Development (INDIACom) 2014 IEEE Xplore March 2014 * Efficient Broadcasting using Preventive Congestion Mechanism in Mobile ad Hoc Network SSubburamm P Sheik AbdulKhader European Journal of Scientific Research 83 2 2012 * A Fault Tolerant Congestion Aware Routing Protocol for Mobile Ad hoc Networks GRajkumar KDuraiswamy Journal of Computer Science 8 5 2012 * Cross Layer based Congestion Control Scheme for Mobile Ad hoc Networks VRamachandra SPujeri Sheeja International Journal of Computer Applications 67 9 2013