Distributed scheduling and power control for cognitive spatial-reuse TDMA networks

We investigate the problem of distributed scheduling and power control for vertical spectrum sharing in spatial-reuse time division multiple access (STDMA) networks. The objective is to minimize the transmission length (in term of time slots) of secondary users (e.g. users in femtocell networks) subject to the interference-limit constraint for primary users (e.g. users in cellular networks) and quality-of-service (QoS) guarantee of secondary users. This problem is known to be NP-complete. We therefore propose a novel distributed two-stage algorithm based on the distributed column generation method to find the near-optimal solution for the transmission schedule. In the first stage, the dual problem corresponding to the transmission length minimization problem subject to the minimum bandwidth requirement of secondary users, called the restricted master problem, is solved to obtain a dual optimal solution at each secondary transmitter. The dual optimal variables are passed to the second stage to solve the pricing problem. The pricing problem here finds a feasible channel access pattern such that the sum of dual optimal variables is greater than 1 subject to the interference constraints for primary users and the signal-to-interference-plus-noise ratio (SINR) constraints for secondary users so that the solution of the master restricted problem can be improved. We also develop a distributed algorithm for solving the pricing problem based on local measurement at each secondary transmitter and a limited number of message exchanges. The proposed algorithm is compared with previously proposed methods and is evaluated in terms of the schedule length and the number of message exchanges.