International Journal of Wired and Wireless Communications Vol.3, Issue 2, 2015

Congestion Control Mechanism in Wireless Network Himanshu Chanyal1, Pranavi Tayal2 , HardwariLal Mandoria3 Department of Information Technology ,College of Technology, GBPUA&T, Pantnagar 1

[email protected] 2

[email protected]

Abstract— One of the major requirements of any network is high speed and reliable communication. TCP itself provides such high speed communication over the dedicated channel to perform the dedicated and reliable connection between server nodes. But the TCP network suffers from the problem of high congestion and the broken link. In such network, if some link loss occurs or some congestion occur. The network gives high degree of data loss. To recover from this loss, there is the requirement of some network reconfiguration approach. The TCP network itself provides the solution of such problem up to some extent. Even then it also suffers from the problem of some data loss over the network. The presented work is in same direction to provide an effective communication so that the data loss will be minimized over the network. In this work a dynamic network reconfiguration approach is defined to analyse the neighbour nodes for the congestion as well as for free bandwidth. In this work, we have presented each node as a monitor as well as router. Each node will monitor the previous node for the congestion analysis and will analyse the next neighbour to find the next possible route for the communication. The presented work has provided the effective communication by performing the reliable reconfiguration of network. Another direction considered in this work is about the analysis of transmitting frames. According to the frame types the size of communication window is defined. The observation is performed on the traffic coming from back nodes and based on this analysis the size of the communication window is decided. The work is implemented in ns2 and the obtained results show that presented approach has provided effective route reconstruction so that the network delay and the loss are reduced. Keywords—TCP, Network.

TCP

congestion

control,

Wireless

I. INTRODUCTION AND PROBLEM STATEMENT One of the core protocols of the internet protocol suite is the transmission control protocol (TCP). It is one of the two

Print-ISSN: 2319-9512 e-ISSN: 2319-9520

original components of the suits, complementing the IP, that’s why the entire suits is commonly referred to as TCP/IP. TCP provides reliable, ordered delivery of a stream of bytes from a program on one computer to another program on computer. TCP suppose that the lost packet due to network congestion, this is reasonable for wired network. But is not suitable for the error prone especially wireless network which has many characteristics like high bit error rate and how bandwidth etc. packet loss may happen due to weather conditions obstacles, multipath interference, mobility of wireless end-devices, signal attenuation and fading etc. when TCP operating in wireless network and wired network. The performance degradation is mainly caused by TCP’s basic assumption that any packet loss is an indication of congestion only, the assumption is not appropriate for wireless transmission error. The appropriate behavior of TCP for the packet loss due to wireless transmission errors is just to retransmit the lost packet without reducing its sending rate. Unfortunately, TCP consider every packet loss as congestion signal, & unnecessarily decrease its sending rate by halving its congestion window size. To avoid such performance degradation, it is important for TCP congestion control mechanism have been presented. The essential of this improved scheme is to distinguish the lost packet due to congestion to those due to corruption. The improved scheme use the congestion control mechanism while the network is congestion and hold the packet sending rate while wireless link corruption. II. TCP CONGESTION CONTROL Above the IP layer, we have the transport layer, where the transmission control protocol (TCP) is the protocol of primary interest. TCP is responsible for driving a data stream into packets, ensure reliable delivery even when the IP layer loses, reorder, or duplicates packets, and at the same time it seems the state of the network to avoid overloading it. The dominating transport layer protocol is TCP. It is used for all kinds of data streams. A TCP connection is a bidirectional, flow controlled, and reliable stream of data between two endpoints, identified by IP address and port number. Our

8

International Journal of Wired and Wireless Communications Vol.3, Issue 2, 2015

primary interest is in TCP connection for transfer of smaller or larger files. For this TCP, it is desirable to get as much data as possible through the network, while at the same time we avoid overloading the network, and share available in a fair way with other users. TCP uses a sliding window flow control. The window limits the amount of data that can be sent without waiting for acknowledgement (ACK) from the receiver.

The objective of congestion is to keep the load of network close to the available capacity. The TCP protocol developed in the late 1970’s resulting in the internet standard RFC 793. The principal for TCP congestion control were developed few years later in response to experience of “congestion collapses” in the internet.

When the window is constant, this result in the so called “ACK clock”, the timing of each sent packet is determined by the reception of the ACK for an earlier packet. One can think about the sliding window and the ACK clock as a peculiar inner control loop which determines the sending rate, when the roundtrip time fluctuates, the sliding window gives an average sending rate of one full window per average roundtrip time. The window size is adjusted depending on received ACK’s and it is details of this outer loop that differ between TCP variants.

The most important concept in TCP congestion control is that of the congestion window. The window is the amount of data that has been sent, but for which no acknowledgment has yet been received. A constant congestion window means that one new packet is transmitted for each ACK that is received. The sending rate is control indirectly by adjusting the congestion size, usually refers to as TCP RENO. To other common variants are TCP New Reno and TCP with selective ACK(Sack).

The objective of the TCP window control is to get a high throughput, close to the connection’s fair share of the available, and at the same time avoid overloading the network. The fair share can vary due to varying amounts of competing cross traffic, and also due to network changes such as to routing updates or wireless links with time-varying capacity. A. CONGESTION CONTROL IN WIRELESS NETWORKS Computer network form an essential substrate for a variety of distributed application, but they are expensive to build and operate. This make its important to optimize their performance, so that users can derive most benefit at the least cost. Though most network perform well when highly used, problem can appears when the network load increase loosely speaking congestion refers to a loss of network performance when a network is heavily loaded. Since congestion phenomenon, can cause data lose, large delays in data transmission and large variation in this delays controlling or avoiding, congestion is critical problem in network management and design. Since these networks carry traffic of a single type and traffic behavior is well known, it is possible to avoid congestion simply by reserving enough resources at the start of each call. By limiting the total number of user each admitted call can be guaranteed to have enough resources to achieve its performance target, and so there is no congestion however resources can be severally under-utilized. Since the resource blocked by a call, even if ideal, are not available to other call.

Print-ISSN: 2319-9512 e-ISSN: 2319-9520

a)

WINDOW BASED CONTROL

b) ACKNOWLEDGMENT AND LOSE DETECTION At the receiving end acknowledgment packet are send in response to received data packets. TCP uses cumulative acknowledgement: each ack includes a sequence number that says that all packets up to that one have been received. Equivalently, the acknowledgement identifies the next packet that the receiver expects to see. When packets are received out of order, each received packet results in an acknowledgement, but they will identify the largest sequence number such that all packets up to that number have been received.eg, if packets 1,2,3,4, and 5 are received, four acknowledgement are generated. The first says “I got the packet #1, I expect packet #2 next”, while the next three acks all say”I got packet #2, I expect packet #3”. The last two acks are duplicate acks, since they are identical to some earlier acks. On the sending side there are several possible reason why duplicate acks are received. Packet delivered by the network out of order, packet dropped by the network, and ack packet duplicate by the network. Packet losses are detected by sender in two ways. Time Out- if packet is transmitted and no ack for the packet is received within retransmission timeout interval(RTO), the packet is considered lost. Fast Retransmission- if three duplicate ack are received, the “next expected packet” from these acks is considered lost. Note that this cannot happen if the congestion window is smaller than 4 packets. Packets that are lost, as detected by either of these mechanisms are retransmitted. The value for RTO is not constant but based on measured average and

9

International Journal of Wired and Wireless Communications Vol.3, Issue 2, 2015

variation of the RTT. It is also modified by the exponential back off mechanism. III. TCP CONGESTION CONTROL STATE There are four distinctive states in the TCP congestion control, illustrated in Figure 2.2, and two state variables related to congestion control: The congestion window cwnd and the slow start threshold ssthresh. The value of ssthresh determines the window increase rule; if cwndssthresh, in which case TCP enters the congestion avoidance state, or • A timeout occurs, in which case TCP enters the exponential back-off state, or • Three duplicate acks are received, in which case TCP enters the fast recovery state. The motivation for the slow start state is that when a new flow enters the network, and there is a bottleneck link along the path, then the old flows sharing that link need some time to react and slow down before there is room for the new flow to send at full speed.

B. CONGESTION AVOIDANCE In the congestion avoidance state, cwnd is increased by one packet per RTT (if cwnd reaches the maximum value,

Print-ISSN: 2319-9512 e-ISSN: 2319-9520

it stays there). This corresponds to a linear increase in the sending rate. On timeout, TCP enters the exponential back-off state, and on three duplicate acks, it enters the fast recovery state. The motivation for this congestion avoidance mechanism is that since TCP does not know the available capacity, it has to probe the network to see at how high a rate data can get through. Aggressive probing would make the system unstable, and a single packet increase seems to work well in practice. C. EXPONENTIAL BACK OFF TCP enters the exponential back-off mode after timeout. Several actions are taken when entering this state: • The lost packet is retransmitted. • The state variables are updated by ssthresh← cwnd/2, cwnd← 1 packet. • The rto value is doubled. When an ack for the retransmitted packet is received, TCP enters the slow start phase. The above update of ssthresh implies that, in the absence of further packet losses, TCP will switch from the slow start phase to the congestion avoidance phaseonce cwnd is increased to half its value before the timeout. If the retransmission timer expires again with no ack for the retransmitted packet, the packet is repeatedly retransmitted, rto is doubled, and ssthresh is set to 1 packet. The upper bound for the rto is on the order of one or a few minutes. Exponential back-off continues until an acknowledgement for the packet is received, in which case TCP enters the slow start phase, or the TCP stack or application gives up and closes the connection. The motivation for the exponential back-off mechanism is that timeouts, in particular repeated timeouts, are a sign of severe network congestion. In order to avoid congestion collapse, the load on the network must be decreased considerably and repeatedly, until it reaches a level with a reasonably small packet loss probability. D. FAST RECOVERY TCP enters the fast recovery state after it detects three duplicate acks. When entering this mode, the first actions of TCP is to retransmit the lost packet, and set ssthresh← cwnd/2. The reason for the ssthresh update is arrange so that the later window increase from cwnd/2 and up will use the additive increase of congestion avoidance, not slow start. TCP then continues to send new data at approximately the same rate, one new packet of data for each received duplicate ack. In rfc 2581, this is described using a fairly complex procedure that artificially inflates cwnd. If no

10

International Journal of Wired and Wireless Communications Vol.3, Issue 2, 2015

ack for the retransmitted packet is received within the rto interval, TCP enters the exponential back-off state. Otherwise, when an ack for the retransmitted packet is finally received, TCP sets cwnd = ssthresh, i.e., half the cwnd value at the start of the recovery procedure, and enters the congestion avoidance state. If more than one packet is lost within the same window, the original fast recovery procedure of TCP Reno is limited in that it can recover only one packet per RTT. This is the main problem addressed by both TCP New Reno and TCP sack. The motivation for the fast recovery mechanism is that the reception of duplicate acks indicates that the network is able to deliver new data to the receiver. PROBLEM IDENTIFICATION Unfortunately present TCP mechanism could not give expected result due to packet loss due to corruption in channel. We know internet technology is changing fast and wireless network is becoming pervasive all around and packet lost in wireless network may not imply congestion in network because it may happen due to weather condition, obstacles and multipath interference, mobility of wireless end-devices signal attenuation and fading. So our assumption upon which network congestion window has been changing in TCP congestion mechanism is not appropriate for wireless. The overall goal of congestion control is to optimize the performance in a communication network. This optimization means, roughly that sending rates at the data source should be as high as possible, without overloading the network. The primary measure of network overload is packet losses.

IV. PROPOSED MECHANISM A. METHODOLOGY This study, focus on the implements a window based flow control mechanism. Roughly speaking, a window based Architecture to flow the data in their form of blocks. The window size is defined as the number of data values transferred at once. In this system flow control was governed simply by the maximum allowed window size advertised by the receiver and the policy that allowed the sender to send new packets only after receiving the acknowledgement for the previous packet. Steps of proposed VoIP based model are as follows: 1. 2.

Define the Source Node S and Designation Node D Convert the Video file to trace file to perform the changing size and rate based communication.

Print-ISSN: 2319-9512 e-ISSN: 2319-9520

3.

4. 5. 6. 7. B.

Convert the network type to WaveLan so that the effective bandwidth utilization and the efficient communication will be performed. Divide the video in smaller fragment to perform the multipath communication. Extract the video frame from the segment Take the routing Decision based on the Frame type Perform the network analysis in terms of number of packets, data lost etc. PHASES OF PROPOSED MECHANISM

The main concept of proposed work is divided in two phases  

Monitoring and Re-Routing

Monitoring is here defined as an analysis approach that is implemented on each node to analyze the node under different parameters. The parameters included in this work are the delay analysis, loss rate analysis and the communication rate analysis. The monitoring process is performed on the backward nodes. As the heavy congestion are detected on the previous node. The route re-configuration will be performed by using the routing approach. This routing approach provides the load distribution as well as the route distribution scheme based on the forwarding node analysis. All the forwarding nodes having the high throughput will be used for the next node communication over the network. The analysis on forwarding node will decide the route generation.In this proposed mechanism, we focus on monitoring and re-routing in order to reduce the congestion. C. SYSTEM ARCHITECTURE The proposed work is simulated in the form of client server and intermediate router. Here the problem is discussed under the effect of modulation scheme. It gives the simulation how the modulation parameter itself is effective to solve the problem of bottle neck as well the problem of bad node over the network. The presented system will improve the throughput in case of congestion and any noise over the network. The system gives the effective use of available bandwidth. After the occurrence of the so called congestion collapse the network. Because of this some congestion detection and avoidance approach is required. Here the bandwidth optimization along with modulation scheme is

11

International Journal of Wired and Wireless Communications Vol.3, Issue 2, 2015

suggested to get the maximum utilization of the bandwidth and to get the maximum throughput over the network. The work also includes the concept of fault detection over the network and the implementation of respective modulation scheme to over the fault loss over the network. The congestion control introduced two new variables for the connection: the congestion window, cwnd, and the slowstart threshold, ssthresh. The window size of the sender, w, was defined to be: w = min(cwnd, awnd), instead of being equal to awnd. The congestion window can be thought of as being a counterpart to advertised window. Whereas awnd is used to prevent the sender from overrunning the resources of the receiver, the purpose of cwnd is to prevent the sender from sending more data than the network can accommodate in the current load conditions. The idea is to modify cwnd adaptively, to reflect the current load of the network. In practice, this is done through detection of lost packets. A packet loss can basically be detected either via a time-out mechanism or via duplicate ACKs. Timeouts: A timer is associated with each packet. If it expires, timeout occurs, and the packet is retransmitted. It will perform an analysis whether the data is been transferred before the timeout. It case of delayed transmission it signify some fault occurred over the network. Duplicate ACKs: If a packet has been lost, the receiver keeps sending acknowledgements but does not modify the sequence number field in the ACK packets. When the sender observes several ACKs acknowledging the same packet, it concludes that the next packet has been lost.

V. WORKING OF THE PROPOSED MECHANISM To understand the whole concept the design is presented as:

Establish the Wireless Mobile Network Setup Each Node as Intelligent Node and keep the statistics of neighboring node Each Node Act as Monitor Node to Detect Congestion over the Previous Node Each Node will act as the router to identify the next neighbors on with route can be distributed to balance the congestion Perform the route distribution to perform the load balancing Analysis of Result under different parameters Fig1.Flowchart of Proposed Mechanism VI. RESULT AND DISCUSSION A. OUTPUT XGRAPHS ON 10 NODES a) PACKET TRANSMISSION VS TIME

Slow start and Congestion Avoidance In slow start, when a connection is established, the value of cwnd is first set to 1 packet, and ssthresh is set to 65535. After each received ACK the value is updated to

cwnd  2  cwnd . The window is increased exponentially until it equals ssthresh, the estimate for the available bandwidth in the network. At this point, the connection goes to congestion avoidance phase where the value of cwnd is increased less aggressively, with the pattern cwnd = cwnd + 1, implying linear, instead of exponential, growth. This linear increase will continue until a packet loss is detected. A packet loss is detected either by receiving duplicate ACKs, or by a timeout.

Print-ISSN: 2319-9512 e-ISSN: 2319-9520

12

International Journal of Wired and Wireless Communications Vol.3, Issue 2, 2015

Figure 2: Packet Transmitted Analysis (Existing Vs. Proposed Approach)

Figure 4: Bytes transmitted Analysis (Existing Vs. Proposed Approach)

Here figure 2 is showing the comparative analysis of packet transmitted over the network. Here x axis represents the time and y axis represents the packet transmitted. As we can see after implementing the proposed approach the successful packet transmission over the network is increased.

Here figure 4 is showing the comparative analysis of bytes transmitted over the network. Here x axis represents the time and y axis represents the bytes transmitted. As we can see after implementing the proposed approach the bytes transmitted over the network is increased.

b) LAST PACKET TIME VS TIME

Figure 3: Last Packet Time Analysis (Existing Vs. Proposed) Here figure 3 is showing the comparative analysis of Last Packet Time over the network. Here x axis represents the time and y axis represents the Last Packet Time of communication. As we can see after implementing the proposed approach the Last Packet Time over the network is increased. c)

BYTES TRANSFERRED VS TIME

Print-ISSN: 2319-9512 e-ISSN: 2319-9520

d) BIT RATE VS TIME

Figure 5: Bitrate (Existing Vs. Proposed Approach) Here figure 5 is showing the comparative analysis of bitrate over the network. Here x axis represents the time and y axis represents the bitrate of communication. As we can see after implementing the proposed approach the bitrate over the network is increased. e)

PACKET DELAY VS TIME

13

International Journal of Wired and Wireless Communications Vol.3, Issue 2, 2015

Figure 6: Packet Delay Analysis (Existing Vs. Proposed Approach)

described the multimedia streaming and all its requirements. Then we various issues on multimedia streaming over TCP have been discussed. Different TCP based multimedia streaming techniques are described with.

Here figure 6 is showing the comparative analysis of Packet Delay over the network. Here x axis represents the time and y axis represents the Packet Delay of communication. Performances of multimedia streaming over single path and As we can see after implementing the proposed approach the multipath TCP have been analyzed through extensive simulation. First streaming over regular single path TCP has Packet Delay over the network is Decreased been analyzed. Then streaming over multipath has been examined. For multipath TCP both homogeneous as well as f) PACKET LOSSRATE VS TIME heterogeneous paths have been considered. From the simulation results we can say that performance of multimedia streaming over multipath TCP improves over single path TCP. In this present work, a route reconfiguration scheme is suggested over the TCP network. To present the work, a wireless network is defined and performed a heavy video data transmission over the network. As the communication is perform over the network using existing network. It performed the route selection dynamically but given the data loss over the network when the bottle neck problem occurred. But in this present work, we have used the route reconfiguration scheme to resolve the problem in case of heavy traffic over the network. Each node of the network is configured as a node as well as the monitor. This monitor node tracks its next as well the previous node for the communication. The Tracking is performed respective to the data loss and the data rate on the immediate neighboring nodes. As the heavy load found, the load distribution is Figure 7: Packet Loss rate Analysis (Existing Vs. Proposed performed from that centralized node. The work will generate Approach) the multiple distributed paths to perform the communication. Here figure 7 is showing the comparative analysis of The obtained results show that the proposed load distribution Packet Loss rate over the network. Here x axis represents the and reconfiguration scheme has improved the network time and y axis represents the Packet Loss rate of throughput and reduce the data loss over the network. communication. As we can see after implementing the proposed approach the Packet Loss rate over the network is Decreased

VII.

VARIOUS PERFORMANCE MATRICES USED IN CONGESTION CONTROL MECHANISM: a) b) c) d) e) f)

The packet loss rate The Packet delay The Bitrate The Packet Transmitted No. of bytes The Last Packet Time

REFERENCES [1] G. Holland and N. Vaidya, “Analysis of TCP performance over mobile ad hoc networks,” in Proc.ACM/IEEE Int. Conf. on Mobile Computing, Seattle, WA, USA, Sept. 1999, pp. 219–230. [2] W. G. Zeng and Lj. Trajkovic, “TCP packet control for wireless networks,” in Proc. IEEE Int. Conf. on Wirelessand Mobile Computing, Networking and Communications (WiMob 2005), Montreal, Canada, Aug. 2005, vol. 2, pp. 196–203. [3] J. Postel, “Transmission control protocol,” RFC 793, Sept.1981.

CONCLUSION In this paper, multimedia streaming over single path and multipath TCP has been explored and analyzed. We have

Print-ISSN: 2319-9512 e-ISSN: 2319-9520

[4] H. Balakrishna, V. N. Padmanabhan, S. Srinivasan, and R. H. Katz, A Comparison of Mechanisms for Improving

14

International Journal of Wired and Wireless Communications Vol.3, Issue 2, 2015

TCP Performance over Wireless Links, ACM SIGCOMM, August 1996. [5] S. H. Low, F. Paganini, and J. C. Doyle Internet Congestion Control, IEEE Control Systems Magazine, Feb. 2002. [6] UmeshD.Nimbhorkar, MohitGarg, “Congestion control in wireless networks”, May 5, 2004. [7] MukeshKumarDhariwal, Sanjeev Sharma,” An improved mechanism for congestion control in TCP for Adhocnetwork”,IJCA(0975-8887) Volume 20, April 2011. [8]

Xu Chang-Biao, Long Ke-Ping, Yang Shi Zhong.Corruption-based TCP rate adjustment in wireless networks. In: Chinese Journal of Computers, 2002, 25(4): pp.438-444.

[9] R. Rejaie, M. Handley, and D. Estrin, “RAP: An end-toend rate-based congestion control mechanism form real time streams in the internet,” in INFOCOM (3), 1999, pp. 1337–1345. [10] Vicente E. Mujica V., DorghamSisalem, Vicente E. Mujica V., DorghamSisalem, “TCP-Friendly Congestion Control over Wireless Networks”, in Proc. Of European Wireless 2004, Barcelona, Spain, February 2004. [11] Yao-Nan Lien and Ho-Cheng Hsiao, “A New TCP Congestion Control Mechanism over Wireless Ad Hoc Networks by Router-Assisted Approach”. [12] Ghassan A. Abed, Mahamod Ismail and KasmiranJumari, “A Survey on performance of congestion control mechanisms for standards TCP versions”, Australian Journal of Basic and Applied Sciences, 5(12): 1345-1352, 2011 ISSN 1991-8178.

Print-ISSN: 2319-9512 e-ISSN: 2319-9520

15