Towards Fair and Efficient Spectrum Sharing Between LTE and WiFi in Unlicensed Bands: Fairness-Constrained Throughput Maximization

The deployment of Long Term Evolution (LTE) networks in unlicensed spectrum is a promising solution to overcome the scarcity of licensed spectrum. Yet it has been widely observed that severe unfairness and performance degradation would occur when LTE coexists with WiFi, the incumbent user of unlicensed bands, without proper adjustment. Fair and efficient coexistence of these two networks thus becomes crucial. It, nevertheless, remains largely unknown how to optimize the total throughput of the LTE and WiFi networks under fairness constraints. To address the above open issue, this paper considers that a WiFi network coexists with an LTE network using the Category 3 or Category 4 Listen-Before-Talk (LBT) mechanism, and aims to characterize the maximum total throughput of the LTE and WiFi networks under two fairness constraints including throughput fairness and 3GPP fairness. The analysis shows that the maximum total throughput is independent of which LBT mechanism the LTE network adopts, and can be improved as the mean successful transmission time of the LTE network increases. Explicit expressions of the optimal initial backoff window sizes to achieve the maximum total throughput under both throughput fairness and 3GPP fairness are also derived, which shed important light on the practical network design. It is found that the initial backoff window size of the LTE network should be enlarged as the mean successful transmission time of the LTE network increases, indicating that for fair and efficient coexistence with a WiFi network, the LTE network needs to access the unlicensed channel infrequently with large packets. To facilitate implementation in practice, distributed schemes are further proposed, with which WiFi and LTE can optimally adjust the backoff window sizes based on their own observation and estimation without the need of coordination between these two networks.