Deep sensing for future 5G communications with mobile primary users
Bin Li, Yijiang Nan, Chenglin Zhao, Arumugam Nallanathan · 2015
A promising joint estimation paradigm, namely deep sensing, is proposed for more challenging spectrum-location awareness 5G applications. A major innovation of the new sensing algorithm is that the mutual interruption between two unknown quantities, i.e. unknown primary states and its moving locations, is fully considered. A unified system model is formulated relying on the dynamic state-space approach, by taking two coupling hidden states into accounts. A random finite set (RFS) inspired Bayesian algorithm is suggested to estimate unknown PU states recursively accompanying its time-varying locations. To avoid the mis-tracking aroused by the intermittent disappearance of PU, an adaptive horizon expanding (AHE) mechanism is designed. Experiments also validate the proposed scheme.