Decentralized Resource Allocation for Multi-Radar Systems Based on Quality of Service Framework

Resource allocation plays a crucial role in the design of multi-radar systems (MRS) for sensing applications. Conventional approaches involve centrally computing the resource allocation solution, assuming the existence of a fusion center (FC). However, these approaches lead to a significant computational burden associated with the FC and fail to yield a viable solution when employing decentralized network architectures. To address the limitations of the centralized approach, this paper proposes a decentralized resource allocation framework. The general resource allocation problem for MRS is comprehensively formulated as an optimization problem based on the quality of service model. To facilitate decentralized optimization, a logarithmic barrier method is employed to approximate the objective function as a linear combination of individual task utility functions. These utility functions can be sequentially updated at each node by communicating with adjacent nodes. The global solution of the optimization problem is obtained when all nodes reach an agreement on resource allocation after a sufficient number of iterations. It is demonstrated that the formulated objective function is unbounded, which is incongruent with the applicable form of common decentralized solution algorithms. To overcome this, a constrained walk alternating direction method of multipliers (CW-ADMM) algorithm is proposed, which ensures an acceptable communication cost while finding the solution. A parallel acceleration approach that employs a broadcast-oriented mechanism is provided to further improve the solution efficiency. Finally, two typical scenarios of MRS resource allocation are investigated to empirically validate the effectiveness of the proposed algorithms.