Performance analysis of data transfer protocols over space communications

The work presented here concerns the study and the investigation of data transmission mechanisms, employed above the network layer of the protocol stack, used to transmit a data file through an end-to-end path composed of low Earth orbit (LEO) and geostationary (GEO) portions. A link built in the unexplored band W connects the Earth station to a LEO satellite (called DAVID). The communication among DAVID and other sites is performed by means of an inter-satellite link in Ka band, which connects the LEO satellite to a GEO one (called ARTEMIS) and the latter to the Earth stations. The main peculiarities of the investigated scenario consist of the visibility of a LEO satellite, limited to a few minutes, and of the only partial availability over time of a return link on the GEO path. LEO visibility may seriously affect the data communication because it implies suspend/resume mechanisms to match nonvisibility periods. The partial unavailability of the return link makes (transmission control protocol) TCP-based solutions barely applicable and implies the design of effective protocol architectures, in order to guarantee a reliable data communication. Several investigations about alternative novel architectures have been produced in order to individuate solutions that meet all the network requirements in terms of service reliability, exploitation of the network resources, and overall service time. The analysis is two-fold: on one hand the attention has been addressed to the LEO path, highlighting the features offered by a transport layer splitting approach and by an application-based solution exploiting the CCSDS (Consultative Committee for Space Data Systems) protocol stack, whose performance is compared with the results provided by the TCP/IP protocol stack, commonly used in the Internet (where IP stands for Internet Protocol). Concerning CEO path, assuming the return link only partially available, two alternatives of the CCSDS protocol stack have been investigated: the first one implies the use of a negative acknowledgement scheme, the second one does not require any availability of a return link. The two paths together compose the end-to-end communication, which is also globally analyzed on the basis of the performance offered by the solutions presented for the two portions. All the results have been obtained by using a simulator developed for the aim.