Abstract :
It has long been understood that Transport Control Protocol (TCP) traffic performs poorly across satellite links. To overcome this performance challenge, SATCOM users have traditionally used Performance Enhancing Proxies (PEP) to "accelerate" TCP traffic. PEPs work by breaking the end-to-end TCP connection, running a different layer 4 protocol across the satellite link, and proxying the TCP connections with each end host. To do this, PEPs must be able to inspect and modify the entire IP packet, including the TCP segment header, to ensure proper handling of TCP traffic over the link. The Department of Defense (DoD) is in the process of implementing the Global Information Grid (GIG), an Internet-like network for warfighters anywhere on Earth. Current plans call for the GIG to transition to a "Black", or IP-encrypted, Core in the future [3]. Within the Black Core, IP packets will be completely encrypted, obscuring all information at layer 3 and above. Such an approach would render existing PEPs useless if placed in the Black Core. This is referred to in this paper as the Black Core problem. Many solutions have been proposed to the Black Core problem. Proposed solutions have included making changes to the network architecture, to the network protocols, or both. This paper expands upon the overview of PEP-based solutions presented in [4], and details several additional approaches. Additionally, this paper overviews the advantages and disadvantages for each method.
