Queue-aware power allocation for multi-way relay networks

We consider a wireless relay network in which multiple single-antenna users communicate with each other through a multi-antenna relay. With our earlier proposed framework of zero-forcing relaying, the inter-user interferences are canceled. In this paper, we investigate power allocation schemes for both regenerative and non-regenerative relays to improve the throughput performance. Considering different traffic demand of each user, a queue-aware power allocation (QPA) scheme is proposed to stabilize the queue of each user by maximizing a “sum demand” metric. The demand metric of each user is proportional to both its channel throughput and its queue length. The fairness in the sense of transmission delay or packet loss ratio is thus achieved as queue length is associated with the two performances. The power allocation problem schemes use Lagrangian method and bisection search. With the QPA scheme, the system throughput performance is improved while keeping all queues relatively stable. The simulation results indicate that the QPA scheme strikes a balance among the fairness of all the users and improve the system throughput.