Abstract:  In a radar-communication integrated network, conventional resource allocation algorithms which aim at maximizing performance in an instant timeslot are not applicable, in the scenario where decisions across timeslots are correlated. This paper investigates a resource allocation decision problem in a radar-communication integrated network, aiming at minimizing the long-term average transmit power consumed by user equipment. The dimensions of the action/state spaces in such a Markov decision process problem grows exponentially as the number of user equipment increases. This leads to “the curse of dimensionality”. To improve computational efficiency of resource allocation, this paper proposes a distributed relative value iteration algorithm. By pre-allocating the resource to each user equipment, the original problem can be decoupled into multiple small-scale subproblems, where each subproblem can be efficiently solved by exploiting a relative-value-iteration-based algorithm. Simulation results reveal that the proposed algorithm can yield better performance than the scheme designed under myopic policies, which merely focus on maximizing the utility in a single slot.