Research On The SCPS Protocol In CCSDS

The CCSDS(Consultative Committee for Space Data System) apply itself to development kinds of communication protocols and data transmission standards in order to meet the space communications environment. For interconnecting TT&C(telemetry,track and command) and Internet, CCSDS launched the SCPS (Space Communication Protocol Specification) as an improvement of TCP/IP. This thesis did the following work around the SCPS protocol stack:First, due to lack of relevant Chinese literature, this thesis detailed study of the SCPS protocol stack. Based on a large number of English document, the SCPS-FP, SCPS-TP, SCPS-SP and SCPS-NP in SCPS protocol stack are analyzed, then the performance of SCPS-FP and SCPS-SP protocol is evaluated preliminarily, and the future of development and application about SCPS-FP and SCPS-SP protocol is discussed.Second, the improvements and expansion of SCPS-TP in response to space communications environment are studied, including how to deal with data loss, bandwidth-constrained, and link asymmetry problem. The extendability of SCPS-TP is studied throughly also.Third, based on RTT prediction, this thesis presents an improvement of SCPS-TP protocol. When using Vegas congestion control algorithm, SCPS-TP protocol need to change congestion window size according to the changed of RTT. Because the topology structure is ever-changing in space network, Vegas can not judge whether the change of RTT is due to congestion or due to the change of the network topology structure. This improvement applied the BP neural network into Vegas, used the neural networks's predictive power to judge the reason of RTT change, thus enhance the adaptability of SCPS-TP. The simulation results show that the program can increase the throughput of SCPS-TP protocol.Then, this thesis presents another SCPS-TP improvement based on classify of network. After successful embed the BP neural network, SCPS-TP can classify the three kinds of network conditions: normal,congestion and topology changes. Then to decide the direction of the congestion window's change. The simulation results show that this program have better performance.Finally, this thesis describes the process of embedded BP neural network into the NS2 simulation platform, and introduces the relevant simulation data handling mechanism.