Cross-layer modeling and analysis of TCP communications over turbulent FSO links

This paper addresses the accurate characterization of the performance of transmission control protocol (TCP) for end-to-end transport services over free-space optical (FSO) links. Unlike previous works on this topic, we introduce a second-order Markovian assumption for the variation of the TCP congestion window to capture the memory from turbulence channels and further propose a Markov chain model that maps two consecutive congestion windows along time into the state space to interpret TCP operations. The analytical expression of steady-state probability distribution vector of the proposed model in general cases where relay routers exist behind the TCP connection is derived, based on which the throughput efficiency of TCP is formulated. The high accuracy of the proposed model is verified by Monte-Carlo simulations and experiments. We further analyze the impact of different link/channel parameters on the TCP performance, and discuss briefly the effectiveness of bundle protocol-based schemes for improving the performance of FSO networks from a transport layer perspective.