Joint Congestion Control And Distributed Scheduling For Wireless Ad Hoc Network

With the continuing development of wireless Ad Hoc networks, many applications require network to support large, bursting data transmission. However, wireless networks bandwidth resource are relatively scarce, and therefore how to achieve the efficient use of resources has attracted wide attention from researchers. Considering the unique features of wireless networks, this paper proposes a channel load-based joint random link scheduling and congestion control cross-layer mechanism, which utilize effectively the wireless bandwidth resource and allocate fairly, under a certain resource constraint.Firstly, we describe the typical layered approach's performance degradation causes, and analyze the existing scheduling and congestion control mechanism to compare their advantages and disadvantages. Then, through analyzing wireless links' competition relation, we elicit the wireless networks' capacity region, and introduce a maximizing aggregate utility-based cross-layer design model. On this basis, we propose a channel load-based joint random link scheduling and congestion control cross-layer mechanism, which consider local "contending clique" as allocation unit "interference region", and use channel utilizing rate to measure offered load, consider region overload as region congestion. According to links' own load and congestion control's virtual packet drop rate, the scheduling mechanism calculate link contending channel probability, and guarantee the probability using starvation alleviating algorithm, provide a greater network capacity and load-based fair basis for rate allocation. Congestion control divided into virtual queue management and source-rate adjustment: the former get interference regions' virtual packet drop rate to measure congestion situation through node channel listening, and let scheduling part achieve virtual drop, in order to alleviate region congestion. The latter combine flows' path virtual packet loss rate with TCP Reno to adjust rate in order to detect available bandwidth or to alleviate congestion.Comparing with existing cross-layer mechanism, the scheduling part adopts more general scheduling sets while considering fair scheduling on link load, elicits more accurate scheduling probability, be closer to optimal capacity region, and make use of starvation alleviating solution replaces impractical synchronization mechanism required generally by these policies to guarantee scheduling probability. Congestion detection part make use of linear function approximation method to estimate congestion signal, which have additive property for reflecting accurately regional congestion. And Congestion control part adopts a reasonable manner, which combines hop-by-hope and end-to-end way, to guarantee long-term stable rate, but also deal with emergency traffic. The resulting mechanism is simple and can be implemented in a distributed in a distributed fashion. The performance of the resulting protocol can be modeled as a utility maximization problem to establish that it indeed leads to a high throughput and fair bandwidth allocation, and the simulation results further prove this conclusion.