Target to sensor allocation means finding trajectories of targets moving over the sensing field and assigning sensors based on them. This is one of the important issues in fields such as military operations, surveillance, biodiversity research for habitat monitoring, and conservation. Yet without proper distributed approaches, the computation, communications, and storage costs will increase drastically. Unfortunately, the sensors in these harsh environments are not designed to reduce these excessive costs. The wrong approach will also make the management of the sensor network much more difficult.

In this paper, the authors propose a new approach for the target to sensor allocation problem in a complex sensor network. The proposed approach is designed using a hierarchical distributed constraint optimization problem (HDCOP). HDCOP divides the initial sensor network DCOP into several DCOP subproblems. These DCOP subproblems are further divided into more subproblems. All DCOPs and DCOP subproblems use nonbinary variables to build a model for the problem. The next level DCOP subproblem will be modeled based on the constraints shared by the previous level DCOP. Given this design, the authors claim that the proposed approach can optimize the objective function of the target to sensor allocation. The objective function also achieves minimum constraint violation.

The authors have properly formalized the problem and set the solution to find a minimum cost assignment of sensors to targets. Since the sensor network is dynamic, the sensors move across the network and can communicate with each other. Using a nonbinary variable can certainly reduce the complexity compared to other methods that use several binary variables per sensor. The authors argue that the complexity (computation) is reduced when the system is partitioned, so hierarchical DCOP will have reduced complexity. Yet they also note that because each level of DCOP will be modeled based on the constraints of the previous level, the approach may increase the communication cost. The authors have not addressed this issue.

The experimental setup is well designed and implemented. The nonbinary modeling has shown significant optimization and produced superior results. However, a theoretical complexity and efficiency analysis would have provided deep insight about the complexity (storage, communication, and computation cost) of the hierarchical DCOP modeling. Other than this, the paper is well written and the proposed methods produce significant results.