Context-Aware Decentralized Invariant Signaling for Opportunistic Communications
A novel scenario-adapted distributed signaling technique in the context of opportunistic communications is presented in this work. Each opportunistic user acquires locally sampled observations from the wireless environment to determine the occupied and available degrees-of-freedom (DoF). Due to sensing errors and locality of observations, a performance loss and inter-system interference arise from subspace uncertainties. Yet, we show that addressing the problem as a total least-squares (TLS) optimization, signaling patterns robust to subspace uncertainties can be designed. Furthermore, given the equivalence of minimum norm and TLS, the latter exhibits the interesting properties of linear predictors. Specifically, the rotationally invariance property is of paramount importance to guarantee the detectability by neighboring nodes. Albeit these advantages, end-to-end subspace uncertainties yield a performance loss that compromises both detectability and wireless environment's performance. To combat the latter, we tackle the distributed identification of the active subspace with and without side information about neighboring nodes' subspaces. An extensive simulation analysis highlights the performance of distributed concurrency schemes to achieve subspace agreement.
READ FULL TEXT