Traffic-aware channel assignment in wireless LANs

Abstract

The importance of channel assignment in wireless LANs has been long recognized and has attracted significant research attention. Despite much recent progress, several important challenges still remain: First, what is the right "performance metric" to optimize for? Most existing approaches to channel assignment try to minimize the number of mutually-interfering APs. However it is not clear if this metric accurately reflects client performance. Second, to what extent is the quality of channel assignment improved by incorporating the observed traffic demands at APs and clients? Recent work in IP traffic engineering has shed light on the tremendous effectiveness of using traffic demands in network engineering decisions. However, to date, no approach to channel assignment has taken traffic demands into account.In this paper, we conduct extensive simulations over publiclyavailable wireless traffic traces, to study the relative efficacy of different optimization metrics in tracking client performance. We show that being "traffic-aware" could substantially improve the overall quality of a channel assignment, irrespective of the metric employed. We also examine the effect of incomplete traffic information on the quality of a channel assignment. Finally, we develop and evaluate a preliminary set of practical traffic-aware assignment algorithms that predict future demands based on historical information and use the predicted demands for assigning channels.

Channel and Interface Management in a Heterogeneous Multi-Channel Multi-Radio Wireless Network

“…Neighborhood RED [14] proposes a variant of the RED algorithm, whereby queues at nodes within two hops are also taken into account, and not just the local queue. Warrier et al have proposed a cross-layer architecture that is based on recent theoretical work on cross-layer optimization [15] Traffic-aware channel assignment in LANs has been considered in [16]. For LANs with uncoordinated access points, it has been proposed in [17], that channel-hopping can help prevent worst-case scenarios, and provide good average case performance.…”

Physical Interference Driven Dynamic Spectrum Management

“…[5], [22] proposed low-complexity algorithms based on local adaptation. [16] proposed algorithms that optimize channel allocation by considering APs' traffic demands. [3] designed centralized algorithms to assign spectrum blocks that are contiguous in frequency.…”

Optimus: SINR-Driven Spectrum Distribution via Constraint Transformation

“…This spectrum distribution problem differs from conventional channel allocation problems because the amount of spectrum assigned to each network node is often unknown and must be determined during the optimization [1]. Solving this problem is an integral part of emerging spectrum related applications including dynamic spectrum auctions [1][2][3][4] and management [5][6][7].…”

“…The majority of existing proposals take an indirect approach: first simplify physical interference conditions into a set of pairwise constraints (or the graph interference model), then distribute spectrum on top of them (or the graph model) [1,[3][4][5][6][7][8]. Such simplification, however, comes at an unpleasant cost.…”