Analysis of ECN on TCP Performance Enhancing Proxy performance for satellite networks

TCP has been a robust and efficient protocol for data communication in the terrestrial wired networks for more than a couple of decades now. However, TCP is very inefficient over high delay and high-error-rate links that might be found in tactical networks especially tactical SATCOM networks. The Space Communications Protocol Standards (SCPS) organization has proposed a more efficient SCPS-TP protocol, which is a modification of TCP to be used for space environments. The Selective Negative Acknowledgement (SNACK) is an important capability provided by SCPS-TP. Because TCP/IP has gone through many incremental refinements, the protocol itself has grown increasingly complex, which makes analytical modeling quite difficult. As a result, much of the evaluation of TCP/IP QoS mechanism has been done using event driven simulators, such as NS-2 or OPNET. More recently, effort for analytical modeling of the performance of TCP has been active. This analysis has been triggered by the limitations inherent in event driven simulations: • Time consuming for network-centric system engineer to use a packet level granularity when investigating the performance of large, complex networks. • Analytical models help network engineer understand the effectiveness of new mechanisms such as TCP performance enhancing proxy and/or ECN being added to the system and implement better requirements and test cases coverage for different network deployment scenarios. SCPS-TP is typically available from Performance Enhancement Proxy (PEP) which serves as a transport layer proxy to split TCP connections. Explicit Congestion Notification (ECN) is used by routers to signal congestion to end points. The analytical model provides quantitative results to compare the ability of ECN signals vs PEP to distinguish between the packet losses due to network congestion and those due to interference or blockages frequently encountered in tactical wireless networks.