Satellite communications are essential to provide Internet access to wide areas, helping bridge the "digital divide". However, long RTTs and the possible presence of losses due to satellite channel errors, severely impair standard TCP performance. To overcome this problem several approaches are possible, including the adoption of enhanced versions of TCP. This paper focuses on these, by presenting the results of a comparative performance evaluation carried out in a satellite environment through a Linux testbed. The interest of the analysis lies in the wide variety of TCP variants considered and in the different aspects analyzed, such as performance at start-up, level of RTT unfairness and robustness against link losses. The results are analyzed at length in the paper and give interesting indications about the performance achievable on satellite channels by the most promising TCP variants proposed in recent years.

Satellite communications pose serious challenges to transport layer performance, mainly because of long propagation delays (especially in geosynchronous systems) and the possi ble presence of random errors on the satellite link. Solutions that cope with these impairments usually rely upon either the adoption of enhanced versions of transport protocols, or the insertion of intermediate agents, like PEPs (Performance Enhancing Proxies). An alternative approach is to adopt the DTN (Delay/Disruption Tolerant Networking) architecture based on the introduction of the new "bundle" layer in the protocol stack. The aim of the paper is to discuss the advantages and disadvantages of the DTN approach compared to the more conventional solutions mentioned. DTN performance is assessed by considering both fully connected networks and networks where continuous connectivity cannot be guaranteed during data transfer and/or at start-up. Performance is evaluated by using the TATPA testbed (Testbed for Advanced Transport Protocols and Architectures), which is based on a cluster of Linux PCs running the DTNperf application, developed to this end by the authors and now included in the official DTN package. Comparative results show that DTN, coupled with TCP Hybla, outperforms NewReno, offering a performance very close to PEPsal and end- to-end Hybla, which are among the most effective solutions on satellite channels. In addition to these encouraging results, DTN offers an intrinsic greater robustness in intermittent or disruptive environments.

The rapidly increasing importance of wireless communications (including satellite) together with the rapid growth of high speed networks pose new challenges to TCP. To overcome them, a wide variety of TCP enhancements has been presented in literature. However, because most proposals aim to address different impairments, the choice of "the best" TCP enhancement becomes arduous, given the increasing level of heterogeneity of present and future networks. The TCP adaptive- selection concept, introduced by the authors in this paper, aims at providing an alternative approach, by challenging the idea that only one TCP enhancement must be adopted. In fact, by extending the concept that underlies ACM (Adaptive Coding and Modulation) to Transport layer, it envisages concurrent adoption of different TCP versions on the same server, the better to match the different impairments present on different connections. Preliminary results obtained by considering a simple application example are encouraging, and justify the following remarks on feasibility and the discussion of some implementation proposals.

The paper describes a measurement campaign recently carried out by University of Bologna (UoB), National Inter-University Consortium for Telecommunications-(CNIT) and University of California Los Angeles (UCLA), on the CNIT GEO Skyplex platform. The aim of the experiments is the performance assessment of a wide range of TCP enhancements on network environments that include a real GEO satellite link. Measurements were carried out by means of the integrated UoB-CNIT testbed, fully controlled by remote through a Web interface. Analysis of live experiments data confirm the challenges posed by GEO sat channels, and show a negative influence of the Skyplex BoD technique on TCP start-up performance. Among the TCP enhancements compared in the paper, best results are provided by TCP Hybla and TCP Westwood, with an advantage of the former in heterogeneous environments characterized by a high level of RTT unfairness.

To overcome the unsatisfactory performance offered by standard TCP variants in the challenging satellite environment, the scientific community has presented a wide variety of possible solutions, encompassing both innovative transport protocols and architectures. Among these, the paper focuses on those TCP variants that have been actually implemented in real operating systems and in Linux in particular. The aim of the paper is to present a sound emulation approach for their performance assessment. This approach relies on the set-up of a complete testbed, able to reproduce, in a fully controlled ("clean") environment, the characteristics of satellites networks, and to provide "live" data from actual implementations of TCP variants. After validating this approach by means of ns-2 results comparison (when available), the paper presents the results of a complete series of tests, which are proposed by the authors as significant elements for a possible future common evaluation benchmark.

Internet communications with paths that include satellite link face some peculiar challenges, due to the presence of a long propagation wireless channel. In this paper, we propose a Performance Enhancing Proxy (PEP) solution, called PEPsal, which is, to the best of the authors' knowledge, the first open source TCP splitting solution for the GNU/Linux operating systems. PEPsal improves the performance of a TCP connection over a satellite channel making use of the TCP Hybla, a TCP enhancement for satellite networks developed by the authors. The objective of the paper is to present and evaluate the PEPsal architecture, by comparing it with end to end TCP variants (NewReno, SACK, Hybla), considering both performance and reliability issues. Performance is evaluated by making use of a testbed set up at the University of Bologna, to study advanced transport protocols and architectures for internet satellite communications.