Randomness and metastability in CDMA paradigms

TL;DR

This paper investigates the properties of sparse random codes in CDMA systems, showing that different models share similar thermodynamic behavior and metastable states, with implications for improving decoding algorithms.

Contribution

It demonstrates the equivalence of thermodynamic results and metastable states in different sparse random CDMA models, and explores analogies to satisfiability problems for better decoding.

Findings

Thermodynamic results are consistent across models.

Metastable states are similar in different code types.

Insights may lead to improved decoding algorithms.

Abstract

Code Division Multiple Access (CDMA) in which the signature code assignment to users contains a random element has recently become a cornerstone of CDMA research. The random element in the construction is particularly attractive in that it provides robustness and flexibility in application, whilst not making significant sacrifices in terms of multiuser efficiency. We present results for sparse random codes of two types, with and without modulation. Simple microscopic consideration on system samples would suggest differences in the phase space of the two models, but we demonstrate that the thermodynamic results and metastable states are equivalent in the minimum bit error rate detector. We analyse marginal properties of interactions and also make analogies to constraint satisfiability problems in order to understand qualitative features of detection and metastable states. This may have consequences for developing algorithmic methods to escape metastable states, thus improving decoding performance.