Deep Sensing for Space-Time Doubly Selective Channels: When a Primary User Is Mobile and the Channel Is Flat Rayleigh Fading
Bin Li, Jia Hou, Xiaofan Li, Yijiang Nan, Arumugam Nallanathan, Chenglin Zhao · IEEE Transactions on Signal Processing · 2016
The unrestrained mobility and dynamic spectrum sharing are considered as two key features of next-generation communications. In this paper, spectrum sensing in mobile scenarios is investigated, which faces still great challenges as both the mobile location of primary-user and fading channel will become time-variant. Such two uncertainties would arouse remarkable fluctuations in the strength of received signals, making most existing sensing schemes invalid. To cope with this exceptional difficulty, a novel paradigm, i.e., deep sensing, is designed, which estimates the time-dependent flat-fading gains and primary-user's mobile positions jointly, at the same time of detecting its emission status. All three hidden states involved by the space-time doubly selective scenario are taken into accounts. A unified dynamic state-space model is established to characterize the dynamic behaviors of unknown states, in which the time-dependent flat fading is modeled as a stochastic discrete-state Markov chain. A Bayesian approach, premised on a formulation of random finite set, is suggested to recursively estimate primary user's unknown states accompanying two others link uncertainties. In order to avoid the mis-tracking of the mobile positions, which is caused either by the incessant disappearance of primary-user or time-variant channels, an adaptive horizon expanding mechanism is also integrated. Numerical simulations validate the proposed scheme.