Abstract
Interference, a main bottleneck in modern wireless communication, may not be always present in many practical situations. In fact, due to the bursty nature of traffic in wireless networks, the corresponding interference is also bursty in general. Such burstiness, if properly exploited, can provide significant gain in performance. To investigate this potential gain, a two-user multiple-input single- output (MISO) bursty interference channel is considered in this work. It is assumed that interference between users is only present with a certain probability. Based on the knowledge of burstiness status, each transmitter adopts a different beamforming strategy and a different rate for communication. Under this setting, we aim to maximize the average system utility and consider the optimal beamforming design for two cases: a) when perfect channel state information (CSI) is assumed, and b) when only channel distribution information (CDI) is assumed at transmitter end. Both optimization problems are nonconvex and difficult to solve. To handle such difficulty, we apply a series of convex approximation techniques such as semidefinite relaxation (SDR) and first-order approximation. Furthermore, we improve the approximation accuracy of our approximation through solving the approximated problem successively, and propose a successive convex approximation (SCA) algorithm. The convergence analysis for the proposed SCA algorithm is also provided. The near-optimal performance of our proposed SCA algorithm is demonstrated by simulations. Our results demonstrate that significant performance gain can be achieved by exploiting the bursty nature of wireless interference network.