In this chapter we performed an evaluation of the most used congestion control protocol, TCP, against a set of new AIMD and rate based congestion control mechanisms. The AIMD approaches used for the evaluation were WCP and TCP-AP. The rate based were XCP, XCP-b and RCP. These mechanisms use network interaction for rate adaptation. The simulations were conducted with the ns-2 simulator. The performance parameters analyzed were throughput, received packets and delay. For the evaluation, it was used various mesh network topologies and several ad-hoc network topologies.
3.5 Conclusions
From the obtained results, it is possible to conclude that WCP overall outperforms the other AIMD based protocols. WCP uses a congestion control mechanism that explicitly reacts to congestion and, it also uses a cooperative communication process between neigh- bor nodes. These processes allow WCP to react more efficiently to the network conditions, making it use more efficiently the medium and the network resources.
TCP-AP results show that this mechanism, as a consequence of using AIMD and rate based congestion control processes, can obtain improved throughput results, when com- pared to TCP, but with less received packets. A TCP-AP sender is using, in whatever type of scenario, an estimate of the current four hop propagation delay, being very conser- vative and becoming more inefficient as the number of hops increases. Another important characteristic of TCP-AP is that it does not use information directly from the MAC layer as the baseline for rate control equation, relying in information from the transport layer, making it react with overall poor performance.
Considering now the rate based approaches, the performed evaluation results show that TCP obtains good results specially when the network is fully utilized and with high mobility scenarios. XCP-b in those scenarios is not taking into consideration packet loss, consid- ering packet loss as a buffer overflow, resulting in an inefficient behavior and introducing slowdown on the network, that are reflected by throughput and delay results.
The results also show surprisingly that TCP has a better behavior than XCP and RCP, being more efficient than XCP and RCP in wireless mesh scenarios. TCP is more fair and stable than XCP and RCP. This is due to the AIMD strategy of TCP. XCP is the less efficient protocol, as it increases delay in the communications. To obtain the available network capacity, both XCP and RCP need that all nodes in the network cooperate, which increases network overhead, specially when dealing with wireless mesh networks. Moreover, TCP, RCP and XCP are not taking into consideration losses due to interference or weak signal strength; this is more relevant in XCP and RCP as they need to evaluate the available capacity. Also, nodes in XCP and RCP are not evaluating precisely network capacity, thus leading to a poor network performance.
However, we know that TCP is not a good congestion control protocol for these net- works, since it does not behave correctly when there are losses due to weak signal strength or interference. In wireless environments, TCP is unable to distinguish between losses due to network congestion or bit errors and handoffs, making TCP-based applications suffering of poor performance. Another problem of TCP is its well known unfairness.
One of XCP and RCP problems is the wrong inference of available bandwidth in wireless networks. For this purpose, cross layer communication may help: MAC layer can be used and be a source of good available rate planning and decision to improve the effective calculation of the available bandwidth of the channel. One possible information is the one obtained by the Network Allocation Vector (NAV). As referred by [47], the NAV is a timer that indicates the amount of time the medium will be reserved. This important information combined, for example, with the times provided by RTS/CTS packets and/or probing packets, can be very important to improve the congestion protocol performance. This cross layer communication mechanism would, then, allow the congestion protocol to decide if it would increase or decrease rate communication, improving throughput and
fairness as bandwidth allocation would also be improved.
Another problem observed in the evaluation was the lack of feedback information in these transport protocols. The interaction of mesh routers to keep track of feedback in- formation, would surely improve XCP and RCP based protocols performance in WMNs. Another point to have in consideration regarding this aspect is that the correct determi- nation of the available bandwidth implies the correct definition of the network achievable capacity. Therefore, queues in all nodes need to be small, leading to more precise feedback information with better channel utilization.
Based on these problems, it is important to define a new wireless inference mechanism that can obtain, without affecting the network dynamics, link capacity and available band- width estimation. The link capacity and available bandwidth results could then be used, through cross layer interaction in real time, by rate based congestion control techniques, like XCP and RCP, to improve congestion control performance and, thus, the overall net- work performance. Available bandwidth and link capacity should be inferred using MAC layer information.
Chapter 4
Real Time Wireless Inference
Mechanism - rt-Winf
4.1
Introduction
Knowledge of link capacity and available bandwidth is important for wireless network design, management and, of course, utilization. [6] states that ”The deployment of wireless networks reveals that despite the advances in physical-layer transmission technologies, limited capacity, and consequently available bandwidth, continues to be a major factor that limits the performance of wireless networks and severe congestion collapses are pervasive”. It is then important to have a capacity and available bandwidth estimation mechanism that can actively and effectively, using network coordination and interaction, measure the referred link parameters. Such technique can then be used for more efficient congestion control
As referred in Chapter 2, link capacity and available bandwidth have been widely stud- ied. A recent mechanism, named IdleGap [1], uses information from the Medium Access Control (MAC) layer focusing on its mechanisms in the CSMA-CA scheme of wireless networks. IdleGap takes the Network Allocation Vector (NAV) into consideration, where NAV represents the time other nodes will occupy the medium. IdleGap introduces a new layer between the MAC and the Network Layer, called Idle Module. The introduction of the Idle Module has an important disadvantage, that is the modification of the Open Systems Interconnection (OSI) Model [48] by the introduction of a new sublayer. The Idle Module is triggered by the NAV vector, registering the occupation time of the wireless link allowing to infer the available bandwidth and the rate value, which is obtained from the IEEE 802.11 header. IdleGap uses the pre-defined IEEE 802.11 header Data Rate value, which is not real, resulting in inaccurate estimation values.
Taking into consideration all the previous considerations we propose a new wireless capacity and available bandwidth estimation technique. The proposed algorithm does not influence network dynamics, uses network cooperation and is a real time estimation technique, called rt-Winf. rt-Winf is based in IdleGap, and aims to eliminate its main
problems: it aims to accurately estimate the capacity and available bandwidth.
This chapter is organized as follows. Section 4.2 describes important background in- formation. Then, section 4.3 describes the main rt-Winf algorithm, presenting the two possible functioning variants, with the RTS/CTS handshake (section 4.3.1) and with probe packets (section 4.3.2). Section 4.4 presents an evaluation of rt-Winf through the obtained results with an emulator (section 4.4.1) and the ns-2 simulator (section 4.4.2). This chapter is closed, on section 4.5, with a summary of its main conclusions.