Robert John Castle

Receiver architectures for HomePNA 2.0

The performance of the Home-PNA 2.0 physical layer is investigated using an appropriate channel model to characterize the home phone line network at around 7 MHz. This reveals considerable dispersion which causes intersymbol interference, so a number of equalizer structures are considered to improve the performance for each of the two possible system symbol rates. The two basic equalization techniques which are analyzed are symbol spaced and fractionally spaced decision feedback equalizers (DFEs) trained with a recursive least square (RLS) adaptation algorithm. Computer simulations show that the performance of the system can be greatly improved at the lower rate, but at the higher rate, which is not mandatory, performance can still be significantly degraded, even with a sizeable equalizer.

Channel modeling and system performance for HomePNA 2.0

The new Home Phoneline Networking Alliance (HomePNA) 2.0 system is an attractive proposition for home networking since it potentially allows transmission rates comparable to those of Ethernet, using the existing installed telephone wiring. However, although the transmission standard is designed to be variable rate (variable symbol rate and bits/symbol), not all combinations are mandatory and it is, therefore, instructive to determine the level of receiver complexity required to handle the optional higher rates. An accurate channel model is devised, characterizing the home phone line network around 7 MHz, which allows the performance of the HomePNA 2.0 physical layer to be evaluated. The very good agreement obtained between the measured and modeled channel responses, in both time and frequency domains, enables a high degree of confidence in the simulation results. Due to the considerable dispersion observed in some home phone line channels, a number of equalizer structures are proposed to improve performance in the receiver. These are compared for each of the two possible symbol rates. Computer simulations show that the performance of the system can be greatly improved at the lower rate with a low complexity equalizer, but at the higher rate, which is not mandatory, a much more complex equalizer is required.

An experimental study of OFDM at 5.25 GHz in an office environment

Orthogonal frequency-division multiplexing (OFDM) has received considerable interest as a means for delivering broadband data in highly dispersive wireless channels. Here we present the design, implementation, and performance of an OFDM system operating at 5.25 GHz, which provides 20 Mbits/s in an office environment. From propagation measurements, we describe a design methodology for determining the key parameters of the OFDM system, such as the number of subcarriers and the size of the guard period. The performance of the system is evaluated using experimental data obtained from a typical large open-plan office. Using a transmit power of +7 dBm, the OFDM system is capable of operating in a highly cluttered environment, providing usable error rates up to a range of 20 m. The importance of error control coding is shown, and we empirically derive the degree of correctability required to achieve packet failure rates of <1% from the cumulative distribution of errors per packet.