Optimal Load Balancing in Cloud Radio Access Networks

Cloud radio access network (CRAN) has been seen as an effective means to address the challenges faced by cellular radio networks, such as high capital expenditure and operating expense, high energy consumption and low spectral efficiency. Especially, CRAN has the potential to equip a cellular network with the load-balancing capability to cope with the non-uniformly distributed traffic in the service area. In this paper, we develop an optimal load balancing scheme for CRAN-based cellular systems by employing an infinite optimization technique. A fairness index is defined to measure the load balancing level of the cellular system and monitored by periodically inspecting the load distribution among all cells. When the fairness index is below a warning threshold, we divide the service zone into compact and connected subregions based on an infinite optimization formulation. Each subregion served by a cell has almost equal area and almost equal throughput requirement. To avoid yielding ill-shaped subregion that is difficult to be covered by a practical cell, a penalty term is introduced to the objective function. Then we update the cell association of each user so that the fairness index return to an acceptable level. Numerical experiments show that our proposal can provide performance-guaranteed load balancing for the cellular network with almost no additional operating expense.