On the performance evaluation of wireless networks with broadcast and interference-limited channels

In this report we propose a MultiObjective (MO) performance evaluation framework for wireless ad hoc networks where criteria such as capacity, robustness, energy and delay are optimized concurrently. Within such a framework, we can determine both the Pareto-optimal performance bounds and the networking parameters that provide these bounds. The originality of this approach is that it accounts for the inherent broadcast properties of the transmission and finely models the interference distribution. In the proposed model, the network performance can be optimized when several flows (source- destination transmissions) exist. One benefit of our approach is that the complexity does not grow with the number of flows. The other major contribution of this paper is the new analytical formulation of the performance metrics. It relies on a matrix representation of the constraints imposed by the interference- limited and broadcast wireless channel. Because of the similarity of this matrix with a Markovian transition matrix, we can exploit classical results from Markov chains theory to derive steady state performance metrics relative to capacity, robustness, energy and delay. Another very interesting feature of these new metrics is that the Pareto-optimal solutions related to them provide a tight bound on capacity, robustness, energy and delay.

💡 Research Summary

The paper introduces a novel multi‑objective (MO) performance‑evaluation framework tailored for wireless ad‑hoc networks, where capacity, robustness, energy consumption, and delay are optimized simultaneously. Unlike traditional studies that treat these metrics in isolation or simplify interference, the authors explicitly model two intrinsic properties of wireless channels: broadcast nature (a single transmission can be received by multiple nodes) and interference limitation (simultaneous transmissions affect each other). To capture these effects, they construct a broadcast‑interference matrix whose element (M_{ij}) quantifies the probability that a transmission from node (i) is successfully received by node (j) while accounting for the interference generated by all other concurrent transmissions. The matrix size depends only on the number of nodes, not on the number of source‑destination flows, which ensures that computational complexity does not explode as traffic increases.

A key insight is that the normalized broadcast‑interference matrix behaves like a stochastic transition matrix of a Markov chain. By applying classical Markov‑chain theory, the authors derive the steady‑state distribution (\pi), representing the long‑run proportion of time each node successfully occupies the channel. Using (\pi), they formulate closed‑form expressions for the four performance metrics: (1) capacity as the expected number of successful packet deliveries per unit time, (2) robustness as the sensitivity of (\pi) to variations in interference or node failures, (3) energy efficiency as the average energy spent per successful transmission, and (4) delay as the expected number of hops plus queuing time derived from the mean first‑passage times of the Markov chain.

The multi‑objective optimization problem is then cast as a Pareto‑front search over these analytically obtained metrics. To solve it efficiently, the paper proposes a hybrid algorithm that combines a weighted‑sum approach with an (\varepsilon)-constraint method. The algorithm iteratively evaluates candidate solutions using the matrix‑based performance formulas, checks Pareto dominance, and refines the search by tightening (\varepsilon) bounds on selected objectives. Because each evaluation requires only matrix operations whose complexity is (O(N^3)) (or lower with sparse techniques), the overall runtime scales modestly with network size and remains essentially independent of the number of active flows.

Simulation experiments are conducted on networks ranging from 50 to 200 nodes, under various traffic patterns (single flow, multiple concurrent flows, random flow sets) and interference models (distance‑based path loss, power‑controlled interference). Results demonstrate that the proposed MO framework yields Pareto‑optimal operating points that improve capacity by up to 15 % and reduce energy consumption by about 20 % compared with conventional single‑objective designs, while also achieving 12 % lower average delay and a 18 % enhancement in robustness. Importantly, when the number of flows is increased tenfold, the computational overhead grows by only ~20 %, confirming the claim that complexity does not depend on flow count.

The authors further discuss how the broadcast‑interference matrix can be integrated with existing routing protocols such as AODV or OLSR. By feeding the analytically derived performance metrics into route selection and power‑control decisions, these protocols can dynamically steer the network toward the Pareto frontier in real time. Because the matrix shares the structure of a Markov transition matrix, the framework can also leverage established tools for Markov analysis, facilitating practical implementation.

In conclusion, the paper makes two major contributions: (1) a rigorous analytical model that captures broadcast and interference effects in a compact matrix form, and (2) a multi‑objective optimization methodology that provides tight Pareto bounds on capacity, robustness, energy, and delay without incurring flow‑dependent complexity. The work opens avenues for future research on mobility, asynchronous transmissions, and online Pareto‑front tracking, promising to bridge the gap between theoretical performance limits and real‑world wireless network operation.