Optimal Spectral Allocation in Citizens Broadband Radio Service

This paper introduces a deterministic optimization framework for spectrum resource allocation, primarily focusing on the Citizens Broadband Radio Service (CBRS). The proposed framework aims to maximize spectrum utilization with minimum possible transmit power while ensuring that operational constraints are met for diverse users. We present mixed-integer linear programming (MILP) formulations for different spectrum allocation configurations, including contiguous and non-contiguous channels and uniform and non-uniform power allocation. To overcome the computational challenges encountered in solving MILPs for large-scale networks, we propose a computationally efficient heuristic method called Sequential Resource Allocation for Warm Start (STRAWS) that addresses resource allocation on one channel at a time. Using extensive computational experiments, we establish that, compared to the conventional contiguous model, the non-contiguous, non-uniform configuration demonstrates an average 15% improvement in low-demand and 75% in high-demand scenarios across all desired Signal-to-Noise ratios. The experiments also reveal that leveraging the STRAWS solution to warm start the optimization process enhances the overall solution quality within tight computational time limits. We also quantify the superior performance of STRAWS in several situations involving trade-offs between the number of users and the total network power. Despite a CBRS focus, our approach readily extends to cognitive radio, IoT, and vehicular networks.