High-level capacity performance insights into wireless mesh networking

Abstract

Despite the increasing interest in wireless mesh networking, it is still unclear how well these systems will scale with increasing user density. Therefore, the aim of this paper is to position mesh network performance within the context of the importance of the mesh-to-access link rate ratio and the number of radios per mesh node when operating the mesh under relatively high traffic loads.

Design and capacity performance analysis of wireless mesh network

“…However, useful high-level insights on the best possible upper bound capacity performance and design options can be appreciated by using the chain analysis results after decomposing a mesh into an equivalent number of logical chains [3]. To obtain such high level insights, it is therefore necessary to assume the following:…”

“…The normalized access capacity of each MAP is analyzed as the number of user domains increases. Figure 2 shows the theoretical upper-bound capacity performance limit of multi, dual and single-radio chain topology network established from [3] based on the earlier assumptions for a range of different mesh-to-access link rate ratios (R). The normalized access link capacity, C = 1 means all the user domains along the chain are able to access their respective access points at full theoretical link rate e.g.…”

“…We therefore propose a simplified approach to understand the performance of WMN by analyzing the relationships between three key design options namely 1) the percentage of wired backhaul points, 2) mesh-toaccess link-rate ratio (R) 1 and 3) number of radio interfaces per mesh node including hybrid radio options across different deployment densities. Some high-level insights into the best possible upper bound capacity performance of WMN have been established in our earlier work [3]. To better understand the above relationships, this paper extends the analysis to compare and contrast the performance of these design options across a number of selected deployment scenarios in central west London.…”

“…As discussed in [3], the capacity performance of WMN is generally influenced by a wide range of factors such as MAC protocol, number of radio interfaces per mesh node, number of user domains sharing the same mesh route, co-channel and adjacent channel interferences, routing protocol, network architecture or topology, antenna types, etc. Numerous studies have been carried out to understand the performances and various design parameters from a large scale deployment's perspective [4], [5], [6], etc.…”

“…More MAPs will have to share the same backhaul point if there are less backhaul points in the network. According to [3], the fair bandwidth-sharing assumption used for the chain does not necessarily result in fair sharing within the whole mesh network. In order to achieve that, it is necessary for the user domains to be distributed across the various logical chains in such a way that all of them will operate at the same value of effective access capacity.…”