There is a large body of work on the multi-hop wireless throughput problem, much of it focused on asymptotic bounds under assumptions such as node homogeneity and random communication patterns. This paper seeks to model wireless interference using a generalized framework that would enable the computation of the optimal throughput the wireless network can support for a given workload. The generality of the methodology and the conflict graphs derived from the protocol model and the physical model, which are used to define the conditions for successful wireless transmission, are viewed as the paper’s chief contributions.
Assuming that packet transmissions at individual nodes can be finely controlled and scheduled by an omniscient and omnipotent central entity, key simulation results show the routes derived from the paper’s methodology often yield noticeably better throughput than shortest-path algorithms. No generalized method is given to determine when to stop computations with the assurance of optimal throughput, although data is presented to indicate that convergence is quite good in many scenarios. These results are presented without involving the physical model, and no results are presented that demonstrate links of different capacities due to resource constraints, although the authors assert they have solved such networks.
For the case of mobile networks, an incremental computational approach based on the presented fixed node models is proposed; again, the hardness of the approach in the physical model case is postponed for further work and investigation. This view of optimal throughput in multi-hop wireless networks bears investigation.