Distributed resource allocation techniques for underlay d2d networks

Abstract

Device-to-device (D2D) network consisting of device pairs communicating directly newlinewith each other without passing the user data through the base station (BS), is an newlineessential component of the fifth generation networks. Underlay D2D network, where newlinethe device pairs reuse the frequency resources of the cellular user equipments (CUEs) newlinethat communicate through the BS, promise an increase in spectral efficiency and newlinenetwork capacity along with improvement in energy efficiency and latency due to the newlineproximity of the devices in the D2D pairs. These gains, however, come at the cost of newlinean increase in interferences that arise due to the reuse; and can be effectively attained newlineonly through judicious resource allocation (RA) schemes. In a D2D enabled cellular newlinenetwork, a centralized RA scheme that assumes the availability of the channel state newlineinformation (CSI) between every device pair at the BS is unrealistic. Distributed RA newlineschemes that enable the devices to make RA decisions, with little to no intervention newlinefrom the BS, are hence of paramount importance. In this thesis, such RA schemes newlinewith distinct features are proposed and analyzed. The proposed schemes can be newlineemployed by the D2D pairs to jointly determine the resource blocks (RBs) and newlinetransmit powers in a distributed manner. newlineFirst, two partially distributed schemes based on Fictitious Play (FP) and its newlinevariant, Fading Memory Joint Strategy Fictitious Play with Inertia (FMJSFP-I), are newlineproposed to determine the RB indices and transmit power levels of the D2D pairs that newlinemaximize the sum rate of the underlay D2D network, by modeling the joint RA newlineproblem as an identical interests game. The convergence of the game to a Nash newlineEquilibrium (NE) is established. newline