Self-supervised learning approach for accelerating wireless network optimization
Optimization tasks in wireless networks can be generically formulated as scheduling link subsets (referred to as independent set (IS)) that do not interfere with each other over the same communication channel. Just like graph coloring problems, finding the optimal set of ISs for multi-commodity flows in a wireless network is considered to belong to the class of NP-hard problems which is challenging to solve. We tackle the link scheduling-based optimization problem in a smart way by developing a self-supervised learning framework which follows a two-stage methodology.