Citation

Abstract

Techniques to reduce the effect of mutually induced cyclostationary crosstalk interference among increasing numbers of digital communication systems will allow for increased system density without a loss of reliability.

Two new analytical results have been derived to evaluate the performance of systems in interference-dominated environments. The first result is a determination of the conditions where a linear equalizer receiver may have enough degrees of freedom to completely eliminate all intersymbol interference, co-channel interference and adjacent-channel interference. These conditions incorporate transmitter and receiver bandwidths relative to the symbol rate as well as antenna diversity. The second result is an expression for the minimum mean square performance of a continuous-time infinite-length decision-feedback equalizer in the presence of multiple cyclostationary interferers and additive white noise.

These analytical results were evaluated for a high-speed digital subscriber-line system to show the performance improvements which can occur over the situation where the interference is stationary with the same power spectrum. Linear equalizer performance curves were also added to the comparisons. In addition, the results of adaptive equalizer simulations were incorporated to quantify the effect of the interference cyclostationarity on transversal-filter equalizer implementations. The issues considered were finite precision coefficients and convergence rates of the stochastic gradient algorithm. These results show two important techniques which can provide opportunities for improved equalizer performance by enhancing the cyclostationarity of the interference. The first is by decreasing the misalignment of the phases of the transceiver clocks in the central office transmitters. The second is by using transmitter pulse bandwidths which are wide relative to the symbol rate.

The second application considered was adjacent-channel interference in digital radio. Both new analytical results showed similar predictions regarding increased spectral efficiency by using equalizers with wide bandwidths. This increased spectral efficiency is due to the equalizers being able to suppress the cyclostationarity adjacent-channel interference, even under conditions of wide transmitter bandwidth where the interference completely overlaps the signal carrying the data of interest.