Pete A. Boyer

A simple generalization of the CDMA reverse link pole capacity formula

A formula that computes the maximum number of users supported per base station in a cellular radio network is generalized to consider the frequency reuse number and arbitrary processing gains. The generalization quantifies a cost associated with in-cell interference by accounting for the lack of interference from the desired user on the total interference and by considering the impact of the frequency reuse number on the out-of-cell interference. This interference cost results in an increase in the received Eb/Io relative to FDMA which should be weighted against a reduction in the Eb/Io requirement resulting from using CDMA.

Performance based on selective multipath reception

As wireless data rate requirements increase, multipath delay spread becomes an increasingly significant limitation on the performance of wireless systems. Techniques such as RAKE reception combat time dispersion by combining multipath components. Alternative implementations of RAKE receivers isolate the strongest multipath components and then shift each component to a common timing reference. The optimal timing reference in frequency-selective fading channels remains an open problem. This paper examines the impact pulse shaping and multipath delay spread on both signal-to-noise ratio (SNR) and bit-error rate performance. The receiver being considered achieves symbol synchronization to the strongest multipath component. The performance when synchronization is achieved based on the first multipath component arrival is also found and used to illustrate performance differences. Multipath delay distributions used on the performance calculations are derived from indoor measurements. Pulse shapes considered in the analysis include root-raised cosine, raised cosine, and Gaussian filters. SNR losses are shown to range between 1-6 dB for bit rates of 10 Mb/s. Results show that synchronization of the receiver to the strongest multipath component gives a 1-3 dB advantage over synchronization to the first arriving multipath component.

Turbo decoder SNR estimation with RAKE reception

The joint performance of a turbo decoder and RAKE receiver using the MAP algorithm depends on the accuracy of the channel reliability factor. In a high data rate/low processing gain environment, inherent interference that results from non-idealities of the RAKE receiver complicate the estimation of the channel reliability factor. The combined performance of a turbo decoder and RAKE receiver is analyzed in a time dispersive and time-varying channel with distinct multipath components. Approaches are examined for estimating the channel reliability factor using the limited information that is known by the RAKE receiver. The sensitivity of performance to SNR mismatches is computed. The impact of the processing gain and the number of multipath components on BER performance is analyzed along with the effect of the channel time coherence. By accounting for the non-ideal RAKE interference effects, improvements in the channel reliability factor calculation result in BER performance improvements on the order of 0.5-2 dB.

Computationally efficient modeling of propagation effects for automated wireless engineering

This paper describes the building blocks for a propagation model that is currently used for automated wireless engineering. These building blocks can be used to construct propagation models for cellular, PCS, MMDS/LMDS, and wireless fixed access. The model starts with a basic analytic design to handle the behavior of the path loss as a function of distance. Then other terms are added to account for other effects, such as terrain obstructions, sloping terrain, road orientation, rain attenuation, and the presence of water bodies along the propagation path. The model has been validated in many urban, suburban and rural markets throughout the US. The model achieves an accuracy of 6-8 dB standard deviation in several urban markets with terrain and morphology data at 30 m resolution. Using a mid-range off-the-shelf PC, the model achieves predictions for a 25 /spl times/ 25 km region in less than 20 seconds at this resolution.