Delay Constrained Utility Maximization in Multihop Random Access Networks

Multi-hop random access networks have received much attention due to their distributed nature which facilitates deploying many new applications over the sensor and computer networks. Recently, utility maximization framework is applied in order to optimize performance of such networks, however proposed algorithms result in large transmission delays. In this paper, we will analyze delay in random access multi-hop networks and solve the delay-constrained utility maximization problem. We define the network utility as a combination of rate utility and energy cost functions and solve the following two problems: ‘optimal medium access control with link delay constraint’ and, ‘optimal congestion and contention control with end-to-end delay constraint’. The optimal tradeoff between delay, rate, and energy is achieved for different values of delay constraint and the scaling factors between rate and energy. Eventually linear and super-linear distributed optimization solutions are proposed for each problem and their performance are compared in terms of convergence and complexity.

💡 Research Summary

This paper addresses the critical issue of transmission delay in multi‑hop random‑access wireless networks by formulating and solving a delay‑constrained utility maximization problem. The authors define network utility as a weighted sum of a logarithmic rate utility and a linear energy‑cost term, allowing system designers to balance throughput, fairness, and power consumption through adjustable scaling factors. Two distinct optimization problems are considered.

The first problem focuses on “optimal medium access control (MAC) with per‑link delay constraints.” Each link’s average packet delay must not exceed a pre‑specified bound. By introducing Lagrange multipliers for the delay constraints, the authors transform the non‑convex problem into a convexified Lagrangian. Using Karush‑Kuhn‑Tucker (KKT) conditions they derive closed‑form expressions for the optimal transmission probabilities and rates. A fully distributed algorithm is then proposed: each node updates its local variables based only on information exchanged with immediate neighbors (the current Lagrange multipliers), guaranteeing convergence to the global optimum.

The second problem tackles “optimal congestion and contention control with end‑to‑end delay constraints.” Here the cumulative delay along any multi‑hop path must stay below a given threshold. The authors approximate per‑link delay as an inverse function of the transmission probability and success probability, thereby linearizing the end‑to‑end constraint. A dual decomposition approach yields a set of local update rules for each router, where a path‑price variable (the dual variable) is communicated across the network. This formulation enables each node to adjust its transmission probability to reduce path delay while simultaneously maximizing the overall utility.

Two families of distributed solvers are presented. The first is a linear gradient‑based algorithm that requires only O(1) arithmetic operations per iteration, making it suitable for low‑power sensor nodes but resulting in relatively slow convergence. The second is a super‑linear (quasi‑Newton) algorithm that incorporates an approximate Hessian to accelerate convergence; although each iteration incurs O(N) additional computation, the number of iterations needed drops dramatically (typically by a factor of three to five). Convergence proofs are provided for both schemes, and simulation results confirm that the super‑linear method reaches the optimal solution in far fewer steps.

Extensive simulations on a 50‑node, five‑hop topology explore the trade‑offs among delay bound, throughput, and energy consumption for various weight configurations. The results show that relaxing the delay bound leads to a steep increase in achievable rates and a concurrent reduction in energy cost, while the overall utility peaks when the delay constraint is moderate (approximately half of the maximum feasible delay). Higher weight on the rate term yields higher throughput at the expense of increased power usage; conversely, emphasizing the energy term preserves battery life but caps the achievable rate. The super‑linear algorithm consistently converges within 30–50 iterations, whereas the linear method requires 150–200 iterations to achieve comparable performance.

In conclusion, the paper delivers a comprehensive framework that explicitly incorporates delay constraints into utility‑based network optimization, and it offers practical distributed algorithms with provable convergence. By jointly considering rate, delay, and energy, the work provides valuable insights for the design of latency‑sensitive IoT, sensor, and vehicular ad‑hoc networks. Future research directions include handling non‑stationary traffic, dynamic channel conditions, and multi‑class quality‑of‑service provisioning.