CHANNEL RECONSTRUCTION WITH LIMITED FEEDBACK IN INTELLIGENT SURFACE AIDED COMMUNICATIONS

Intelligent reflecting surface (IRS) has been promoted as a leading candidate technology for enhancing coverage as well as spectral and energy efficiencies in future wireless communication networks. An IRS comprises of a multitude of low-cost antenna elements that can be programmed to influence impinging electromagnetic waves in a desirable manner. However, performance enhancements are conditional upon availability of accurate channel estimates, which are especially hard to obtain for a passive IRS that lacks baseband processing capability. In this work, we propose novel channel reconstruction formulations for IRS-assisted communications where the IRS panel has only passive elements and the intended receiver provides just signal strength feedback reports. Our formulations simultaneously exploit low rank property and sparse beam-space representation of the unknown effective channel, and can accommodate subspace side-information whenever available. We design efficient proximal distance based algorithms to reconstruct the effective channel and demonstrate their superior performance via results generated using the open-source SimRIS platform.