This paper presents a novel solution optimizing satellite access lower layers and system performances called variable size packets (VSPs) designed for the transport of Internet protocol (IP) datagrams or any packet type from the network layers characterized by a highly variable size. VSP objective is to maximize the efficiency of layers 1 and 2 through a joint optimization of access and transport mechanisms mainly for satellite systems addressing residential or corporate terminals market through a multiple-access scheme. It is particularly attractive for the return link of next-generation satellite systems to be IP-based, as a possible evolution of the digital video broadcast-return channel via satellite (DVB-RCS) standard. The VSP solution relies on the definition of multiple encapsulation sizes, and is associated to a waveform and an access adapted to variable size packets' transport. It uses the traffic's variable size nature to increase encapsulation efficiency, coding, and spectral efficiency performances, and system flexibility. It is well adapted to systems using an adaptive waveform leading to system capacity's increase and better performances. This paper describes the VSP principle, advantages and possible high-level system tradeoffs. It then focuses on one critical issue, the coding, and presents a suitable coding scheme's choice with regards to performances and implementation complexity.

Most of GEO satellite telecommunication systems today aim at providing multimedia access services to low cost and power-limited user terminals by implementing performant regenerative onboard processing techniques (demodulation, routing). In order to meet high QoS performance at low SNR, especially on the satellite uplink, performant coding schemes such as turbo codes are necessary. We focus on the critical point of the on-board demodulator: the carrier phase recovery performed in presence of powerful turbo decoding. Non-data aided (NDA) and TD estimators are presented and analysed in terms of characteristics for transmissions using recursive systematic convolutional (RSC) codes that can be used for turbo codes. We show that TD estimation can operate down to E/sub c//N/sub 0/=0 dB with frequency offset up to 10/sup -3/.

In this paper, we describe and analyze a new feedforward carrier-phase algorithm suited to on-board digital implementation. The algorithm studied here (for phase recovery of M-ary phase-shift keying signals) has the particularity to perform all calculations in polar coordinates in order to reduce the implementation complexity. The defined phase estimator is operating at very low signal-to-noise ratio with small frequency offsets. Its performance is close to those of the Viterbi and Viterbi algorithm.