Use of a Weighted Conflict Graph in the Channel Selection Operation for Wi-Fi Networks

Abstract

Allocation algorithms in IEEE 802.11-based WLAN, that consist to associate to each AP a channel, are mainly based on a conflict graph that represents the conflicts (interference, CCA detection, etc.) between APs. In this paper, we propose to use an enriched version of the conflict graph, namely a weighted conflict graph. This latter models the CCA detection that can be total (all transmissions are detected) or partial. Beside, a model is given to compute the throughput of each AP for a given allocation. This model is combined to a greedy algorithm that aims to find the allocation that maximizes the proportional fairness. Simulations based on the recent IEEE 802.11ax amendment are carried out for small WLANs with a few APs and very dense networks. Results show that when the network is dense or constrained, there is a gain to consider partial detection.

Decentralized Channel Management in WLANs with Graph Neural Networks

“…Prior solutions on WLAN channel allocation rely on a central controller to manage APs and make allocation decisions with the global network information [4]- [10]. The problem can be formulated as a weighted graph coloring problem, which is NP-complete.…”

“…The problem can be formulated as a weighted graph coloring problem, which is NP-complete. To tackle the NP-hardness, various schemes, such as local-search [4], integer linear programming [5], deep reinforcement learning (DRL) [8], [9], etc., have been proposed. Among them, three notable related works are TurboCA [7], Net2Seq [8], and graph convolutional network (GCN)-based channel allocation [9].…”