Physical Layer Aware 5G Ran Slicing

Athira M G




1 Research Work

Network slicing in 5G enables a multi-service network that can support a wide range of verticals with specialized service requirements on a single physical network by slicing them into multiple isolated logical networks. 5G RAN slicing is a part of an end-to-end network slicing deployment where the Network Operators (NO) manage Radio Access Network (RAN) resources, needed for network slicing to operate. Due to the increase in demand for high throughput, low latency wireless cellular connections, we must develop dynamic and efficient resource allocation policies to address the issues and guarantee a better user experience. My project goal is to develop refined algorithms to solve the RAN slicing enforcement problem and maximise the total sum rate by doing power allocation over the resource allocation. We plan to revisit the linked PRB allocation problem to account for pathloss and multipath fading. Intuitively, more the SNRs are comparable for the UEs from BSs, the more aligned the PRB allocation should be to a slice in the BSs that serve these cells. This differs from the literature, which only considers whether there is an interference or not. The second goal of our work is to avoid heuristics employed by the algorithms in the literature to increase the number of linked PRBs. Instead, we formulate the linked PRB maximization algorithm as a DC programming to solve this challenging problem. Furthermore, by improving the algorithm’s performance to allocate resources efficiently, we can increase the received SINR by the UEs and thereby increase the net throughput. The final goal of our work is to do power allocation on the PRB allocation which is also an DC programming problem and gives us a suboptimal power allocation in polynomial time. Therefore, by improving the resource allocation algorithm and implementing power allocation on top of it we are able to increase the total sum rate over all PRBs across all BSs.