KRAFT: Kalman Filter based Energy-aware LTE and Wi-Fi Joint Throughput Optimization

Abstract

Resource block allocation in LTE is a complex metric consisting of available bandwidth, required data rate and channel state conditions. A Wi-Fi access point, co-operating with LTE eNB can service a single User Equipment (UE) through multiple interfaces (multihoming) for improving throughput. The challenge is to inter-operate the greedy LTE scheduler with deferral-based Wi-Fi band allocation. A joint scheduling therefore is a complex problem requiring a robust estimation method which not only optimize per-user throughput but also can support a minimum average system throughput condition. Kalman filter provides resource prediction while minimizing the mean square error. In this paper, we propose KRAFT, a novel LTE and Wi-Fi joint scheduler using Kalman filter. We periodically measure the discrete SINR values, available bandwidth and resource requirements to predict the system throughput optimized scheduling among Wi-Fi and LTE networks. Also, an energy-aware scheduling algorithm is proposed as a tie breaker between all the other approaches.