Joint Asynchronous Congestion Control and Distributed Scheduling for Multi-Hop Wireless Networks

Fairness and Throughput Guarantees with Maximal Scheduling in Multi-hop Wireless Networks

“…This comparison characterizes the penalty due to the use of only local information in the scheduling. A common feature of all the existing results in this context has been that same performance bounds are obtained for all sessions [2], [4], [5], [13], [14]. This uniform characterization therefore bounds the performance of the network in terms of that of the worst session.…”

“…al. [2] have shown that in a specific interference model, the node exclusive spectrum sharing model, maximal scheduling can be used for maximizing the network utility and congestion control. One of our important contributions is to show that maxmin fairness can be attained in wireless networks with arbitrary interference models within the framework of maximal scheduling, without sacrificing the simplicity and the distributed nature of these policies.…”

Randomized Algorithms for Throughput-Optimality and Fairness in Wireless Networks

“…We let q n,d [t] denote the length of the queue at node n keeping packets destined for node d at the beginning of slot t. We consider a general interference model formulation that contains all of the graph theoretic collision models considered in the context of scheduling (e.g. NESS [20,14,27,3], or two-hop interference models [28,4]). We say that two links interfere if their concurrent transmissions collide, and assume that if two interfering links are activated in a slot, both of the transmissions fail.…”

“…Theorem 1 shows that under very mild conditions, if a randomized scheduler can be found that satisfies (3), and the control in (4) can be performed, then the stability will be achieved for any stabilizable incoming traffic.…”

“…In particular, [14,27] provided algorithms that guarantee 50% utilization of the stability region for node-exclusive-spectrum-sharing (NESS) interference model, where each feasible allocation forms a matching 1 of the graph. It has been shown in [3] that 33% can be guaranteed even when the algorithm operates in a totally asynchronous manner (see [2] for a definition). Distributed implementations exist for this particular interference model if a significant portion of the capacity is sacrificed.…”