I. Introduction
Low signal-to-noise ratio (SNR) systems are relatively pervasive; consequently, the challenge of successfully detecting the transmitted signal in noisy environments has an axiomatic urgency. Various techniques have been proposed to address low-SNR detection, including coding [1], [2], [3], improved phase detection [4], stochastic resonator [5], energy-based [6], pre-filtering [7], [8], and spread spectrum [9] solutions. These systems report poor BER performance (greater than ) for SNR values much lower than 0 dB. For standard-size packets (∼12 000 bits) with no error correction (only detection, such as checksum) the probability of packet loss in these systems is near 100%. Even recent work [5] in low-SNR link detection struggle to obtain low BERs, demonstrating that this is an important on-going topic. In this work, the proposed system is compared to a coding and a spread spectrum case. Using a discrete integrator technique and a physical-transport cross layer control design, it is possible to achieve high reliability detection under very low-SNR schemes, without relinquishing considerable throughput. The transmission control protocol (TCP) is designed to manage data flow based on congestion assuming losses are extrinsic, nevertheless, it can be configured to provide a dynamic control over physical layer parameters with the objective of reducing intrinsic loss. This technique is best suited for lossy-channel systems with low number of simultaneous TCP transmissions and large round-trip time (RTT). Because the physical layer is unique to this system, it is unlikely it can communicate with standard networking equipment, therefore it is necessary to surround the low-SNR link with TCP proxies to prevent out-of-order segments, as shown in Fig. 1. An additional advantage of implementing TCP proxies is the reduction of the perceived RTT, hence increasing the end-to-end throughput. Employing TCP proxies surrounding low-performance links has been previously employed in satellite networks [10], but not under a cross-layer physical-transport system.