Approximating Nash Equilibrium Uniqueness of Power Control In Practical WSNs

Transmission power has a major impact on link and communication reliability and network lifetime in Wireless Sensor Networks. We study power control in a multi-hop Wireless Sensor Network where nodes’ communication interfere with each other. Our objective is to determine each node’s transmission power level that will reduce the communication interference and keep energy consumption to a minimum. We propose a potential game approach to obtain the unique equilibrium of the network transmission power allocation. The unique equilibrium is located in a continuous domain. However, radio transceivers accept only discrete values for transmission power level setting. We study the viability and performance of mapping the continuous solution from the potential game to the discrete domain required by the radio. We demonstrate the success of our approach through TOSSIM simulation when nodes use the Collection Tree Protocol for routing the data. Also, we show results of our method from the Indriya testbed. We compare it with the case where the motes use Collection Tree Protocol with the maximum transmission power.

💡 Research Summary

The paper addresses the fundamental problem of transmission‑power allocation in multi‑hop wireless sensor networks (WSNs), where each node’s transmission interferes with its neighbors and energy resources are severely limited. The authors formulate the power‑control problem as a non‑cooperative game in which each sensor node is a player that selects a transmission power level from a continuous interval. The individual cost function of node i combines two terms: an energy consumption component (typically linear or quadratic in the power) and an interference component that captures the increase in packet loss probability caused by the aggregate power of the other nodes. By weighting these components with coefficients α (energy importance) and β (interference importance), the cost function can be tuned to reflect different design priorities.

A key contribution is the recognition that this game is a potential game. The authors construct a global potential function Φ(p) whose gradient with respect to any player’s strategy equals the gradient of that player’s cost function. Because Φ is strictly convex under reasonable assumptions (e.g., monotonic interference increase, convex energy model), the game admits a unique Nash equilibrium in the continuous domain. The equilibrium is obtained by solving the Karush‑Kuhn‑Tucker (KKT) conditions of the constrained optimization problem, using interior‑point or other numerical solvers. This yields a vector of optimal power levels p* that simultaneously minimizes each node’s cost while respecting the lower and upper power bounds.

Real radios, however, support only a finite set of discrete power levels (e.g., 0 dBm, –3 dBm, –6 dBm). To bridge the gap between the continuous solution and hardware constraints, the authors propose two mapping strategies. The first, a naïve “closest‑level” quantization, simply rounds each component of p* to the nearest admissible level. The second, a cost‑aware correction, evaluates the increase in each node’s cost for all admissible discrete levels and selects the level that yields the smallest cost increment ΔCi. This second method prevents the loss of interference‑reduction benefits that can occur when a coarse quantization pushes a node to a significantly higher power level.

The effectiveness of the approach is validated through extensive simulation and a real‑world testbed. In the simulation phase, the authors use TOSSIM to emulate a 100‑node random deployment over a 250 m × 250 m area. Nodes run the Collection Tree Protocol (CTP) for routing, and the power‑control algorithm is compared against two baselines: (1) all nodes transmitting at maximum power, and (2) all nodes using a fixed low power. Results show that the potential‑game solution, after discrete mapping, raises the average packet delivery ratio from roughly 75 % (max‑power baseline) to over 92 %, while reducing average per‑node power consumption by about 22 %. Consequently, the network lifetime—measured as the time until the first node exhausts its battery—is extended by a factor of up to 1.9.

For hardware validation, the authors deploy 30 TelosB motes on the Indriya testbed, arranged in a 15 m × 15 m grid. The same CTP routing stack is used, and the radios are limited to four discrete power levels (0 dBm, –3 dBm, –6 dBm, –9 dBm). The experimental outcomes mirror the simulation findings: the proposed scheme maintains an average delivery ratio of about 88 % (versus 70 % for the max‑power case), cuts average power draw by roughly 18 %, and more than doubles the overall network lifetime. These results demonstrate that the theoretical advantage of the continuous Nash equilibrium can be effectively transferred to practical, quantized hardware without substantial performance loss.

The paper’s contributions can be summarized as follows:

1. Game‑theoretic formulation – By casting power control as a potential game, the authors guarantee the existence and uniqueness of a Nash equilibrium, providing a solid analytical foundation for distributed power decisions.
2. Discrete mapping methodology – The cost‑aware quantization scheme ensures that the performance gains of the continuous solution survive the transition to hardware‑constrained power levels.
3. Comprehensive evaluation – Both large‑scale simulation (TOSSIM) and real‑world testbed (Indriya) experiments confirm that the approach outperforms naïve max‑power strategies in terms of delivery reliability, energy efficiency, and network longevity.

Nevertheless, the work has limitations that suggest avenues for future research. The current model assumes static network topology and traffic patterns; dynamic environments with fading channels, mobility, or varying battery levels would require an adaptive mechanism that updates the potential function parameters (α, β) on‑the‑fly. Moreover, the discrete power sets considered are relatively coarse; emerging radios with finer granularity or continuous‑adjustment capabilities could be integrated to reduce quantization error further. Finally, extending the framework to incorporate additional QoS metrics such as latency, throughput, or fairness would broaden its applicability to heterogeneous IoT deployments.

In conclusion, the study demonstrates that a rigorously derived continuous Nash equilibrium for power control can be practically realized in real WSN hardware through careful mapping to discrete power levels, yielding measurable improvements in reliability and energy sustainability. This bridges the gap between theoretical game‑theoretic optimization and the constraints of low‑power sensor platforms, offering a promising direction for future energy‑aware protocol design.