Semidefinite relaxation and randomization for dynamic cell association in heterogeneous networks

In this paper, we consider a heterogeneous network with one macro node and one pico node. We are concerned with the problem of associating users in the macro cell, to either the macro node or the pico node, in order to maximize the sum rate in the downlink. We formulate a new theoretical framework to study this problem and derive an upper bound on the achievable sum rate using semidefinite relaxation. Furthermore, we propose a randomized heuristic to produce a feasible solution, and most importantly, give an analytic guarantee on its performance. Independently of the problem data, we can ensure a worst case performance for the randomization method. In practice, this guarantee is as good as the standard best SNR heuristic typically used in 3GPP LTE networks.