The performance of TCP degrades severely over satellite links, since the throughput of this protocol is inversely proportional to RTT. On one hand, Demand Assigned Multiple Access (DAMA) methods assign users variable time-slots matching user information transmission requirements. Moreover, DAMA-based systems have full control and overview of the system resources. Although these protocols show significant system improvement, they have several drawbacks, e.g. the delay due to the signaling procedure, also the overhead of the signaling messages. Considering an application scenario with light traffic load in uplink, Random Access (RA) schemes are valid approaches to mitigate RTT issues. RA methods show low latencies, especially when they implement Successive Interference Cancellation (SIC). This concept is believed to further reduce the delay by means of recovering from collisions, if one of the collided packets is known.
The aim of this thesis is to analyze in terms of theory and simulation the performance of TCP (mice or elephant connections) over satellite links implementing random access techniques with SIC on the return channel. Towards this end, maximum throughput and stability issues will be investigated. The goal is to show that the delay seen by TCP using these methods is not critical and can improve its performance.
The expected output of this task is a functional implementation (in NS-2) of innovative random access methods. Further, the overall system performance should be analyzed in order to support the simulation results.
Protocol implementation (+Documentation), protocol testing and evaluation.
6 - 8 month
Offer valid till