Semi-Blind Interference Prediction in Wireless Networks

Our research investigates the concept of interference prediction as an unprecedented approach for interference management and medium access in wireless networks. This paper is a first step in this direction: it proposes and evaluates a simple interference prediction technique that is based on low-complexity learning. Nodes predict the interference situation they expect to experience in the near future and select the most favorable time slot to start the transmission of a multislot message. The performance gain is evaluated in a small-scale fading environment in terms of link outage and delay against random slot selection. Simulation results show that interference prediction is a promising building block for wireless systems. Additional studies are needed to explore advanced techniques and assess their feasibility.