User association and load balancing in long term evolution network in Addis Ababa, Ethiopia

The exponential growth in mobile data demand is increasing the need for high capacity provisioning. To satisfy this demand, mobile operators are deploying latest generation of cellular technologies, such as Long Term Evolution. Moreover, the non-uniformity in service demand distribution is creating load imbalance on wireless networks. This imbalance reduces the capacity of the networks and various load balancing strategies are being implemented to circumvent it. This paper investigates a geographic load balancing algorithm, called Bubble Oscillation Algorithm in LTE networks and beyond. The BOA is customized to make it feasible in a real LTE network. A realistic test case network in Addis Ababa, Ethiopia, is used to validate the algorithm and compared it with the default LTE user association criteria, the best Reference Signal Received Power. The numerical results show that, BOA improves users throughput at the cell-edge up to 67.1% as compared to the best RSRP. Considering the number of unassociated users as a second criteria for comparison, service outage for BOA is 7.20% while that of the best RSRP is 40.93% when the service demand is equal to total system capacity, which is a significant improvement.