On Improving Clock Synchronization Accuracy for LTE-A Networks

The need to synchronize wireless networks very accurately is slowly but steadily gaining importance recently among the research community as well as network operational teams. This is due to the rapid advancement of wireless access technologies like the LTE-A, 5G, etc. Within this context, in this paper we propose, design, evaluate a model based reinforcement learning algorithm that aim to improve network clock synchronization accuracy suitable for future networks. The proposed algorithm embraces a new metric which identifies asymmetric links in the backhaul networks in advance and prevents re-routing of timing packets via these links. By this, we claim that link asymmetry delays (LADs) which is a major cause for poor timing accuracy could be eliminated. Experimental results are presented to show the effectiveness of our approach.