Energy efficient downlink scheduling in LTE-Advanced networks

LTE-Advanced (LTE-A) cellular networks are characterized by their use of OFDM based multiple access (OFDMA) and multi-antenna (MIMO) techniques. Starting from the seminal work of Andrews-Zhang, several algorithms for subframe scheduling over such networks, that seek to exploit multi-channel, multi-user and spatial diversity, have been proposed. Recent interest in both industry and academia is to incorporate energy efficiency in the design of such algorithms. In this paper, we formulate a single-cell downlink scheduling problem that exploits all key features and enforces all the main mandatory constraints on the choice of transmission parameters. Linear cost constraints on the set of channels are also imposed in order to accommodate energy efficiency considerations. We show that the resulting problem is equivalent to a combinatorial auction problem with fractionally sub-additive (but not necessarily submodular) valuations and linear constraints on the items being auctioned. The combinatorial auction problem with submodular valuations and without any additional constraints, can be approximately solved using the classical greedy algorithm. For the problem at hand, we instead propose a deterministic multiplicative updates based algorithm. The latter algorithm is simple enough to be implemented in the fine time-scale (1 ms) of subframe scheduling. We rigorously establish the approximation guarantee of our proposed algorithm and evaluate its performance via simulations.