Spatial fairness in linear wireless multi-access networks

Multi-access networks may exhibit severe unfairness in throughput. Recent studies show that this unfairness is due to local differences in the neighborhood structure: Nodes with less neighbors receive better access. We study the unfairness in saturated linear networks, and adapt the multi-access CSMA protocol to remove the unfairness completely, by choosing the activation rates of nodes appropriately as a function of the number of neighbors. We then investigate the consequences of this choice of activation rates on the network-average saturated throughput, and we show that these rates perform well in a non-saturated setting.

💡 Research Summary

The paper investigates the pronounced throughput unfairness that arises in saturated linear wireless multi‑access networks employing CSMA‑type protocols. The authors first model the network as a continuous‑time Markov chain where each state corresponds to an independent set of simultaneously active nodes. Leveraging the well‑known product‑form stationary distribution, they show that when all nodes use the same activation rate, nodes with fewer neighbors (typically at the network edges) obtain a higher probability of transmission, while centrally located nodes suffer reduced access. This spatial bias is identified as the root cause of unfairness.

To eliminate the bias, the authors propose a simple yet powerful rate‑allocation scheme: the activation rate ν_i of node i is set proportional to the number of its neighbors plus one, i.e., ν_i = α·(d_i + 1), where d_i denotes the degree (number of interfering neighbors) and α is a global scaling factor that controls overall load. Substituting this choice into the product‑form expression yields identical marginal activation probabilities for all nodes, regardless of position, thereby achieving perfect fairness (Jain’s index = 1).

The paper then examines the impact of this fairness‑inducing design on network‑wide performance. Because the average number of active nodes depends only on α, the scheme can be tuned to approach the theoretical maximum saturated throughput. Importantly, the authors prove that the fairness‑preserving rates do not significantly degrade the aggregate throughput compared with the uniform‑rate baseline.

Beyond the saturated regime, the authors extend the analysis to a non‑saturated setting where packets arrive according to independent Poisson processes. Simulations demonstrate that the neighbor‑aware rates continue to deliver near‑optimal throughput while maintaining low queue lengths and delays across the network. The scheme is fully distributed: each node requires only local knowledge of its degree, making it readily implementable in dynamic wireless environments where topology may change.

Overall, the study provides a rigorous analytical framework, a practical rate‑selection rule, and extensive performance evaluation, establishing that neighbor‑dependent activation rates can completely eradicate spatial unfairness in linear CSMA networks without sacrificing, and often improving, overall throughput.