Forecasting economic quantities can bring high benefits to business planning and operation. To provide accurate forecasts and to investigate the factors behind the trends, we present a new prediction system in this paper. Applying it to the Number of Mobile Subscribers (NMS) in the United States, China, and Germany, we observe interesting differences between saturated and emerging markets. Moreover, we find that prediction accuracy highly increases when (i) external factors are systematically included in the prediction model, while factors with ambiguous effects are removed and (ii) the time lag between a factor's change and using it for prediction is optimized. By doing both, our system is not only more accurate than common forecasting methods but reveals if and how quickly a factor affects a trend's accuracy.

We investigate various array combining algorithms considering the special requirements of the UTRA/FDD uplink. After formulating the algorithms, we assess their performance in a modified vehicular B channel that is spatially inhomogeneous. The bit error ratios in different situations are presented together with some considerations about the computational complexity of the used methods. From bit error performance and required effort, we conclude that the minimum mean squared error (MMSE) algorithm with reduced complexity is the most suitable method for a real operating system.

We present a robust smart antenna processor that is based on directions-of-arrival (DOA) estimation at the uplink. For optimum BER performance in a cellular mobile radio environment we resolve all relevant paths for the intended user, not only a single one. To increase the system's reliability, we apply separate trackers for uplink and downlink and use a beamforming algorithm that places broad nulls to suppress interferers as they appear realistically in the mobile radio channel. A performance evaluation of the eight-element array based on a semi-stochastic directional channel model shows that the DOA estimation accuracy is not of great concern, but estimation robustness strongly influences the BER. By adaptation within every GSM frame, the processor yields a mean noise and interference ratio (SNIR) improvement of up to 35 dB in an interference limited mobile radio macro-cell environment. The tracking algorithm gains 10 dB in mean input SNIR at a raw BER of 1%. Our beamformer with broad nulls has up to an order of magnitude better BER performance than a conventional beamformer that places sharp nulls. Angular diversity in the uplink and spatial interleaving in the downlink are implemented to improve the link quality.

Adaptive space division multiple access (A-SDMA) is a new technology for mobile communication systems with which the capacity can be increased and cochannel interference can be reduced. It is based on the application of an adaptive array antenna at the base station site which is used to form multiple independent beams per traffic channel. Each of these beams is used to support a connection with a mobile station, i.e., multiple mobile users can be served in the same traffic channel, and the capacity of the base station is increased. To evaluate the benefits of A-SDMA in terms of capacity increase, an analysis is presented which permits to analyse the spatial multiplex gain of an A-SDMA system taking into account the user mobility. Additionally, various other performance parameters of particular interest for the design of a cellular system with A-SDMA technology are derived. Furthermore, an example is given, where from a cellular scenario a user density distribution is derived which is then used as an input for the analysis derived in the first part of the paper.

The author addresses the problem of selecting timer thresholds for the asynchronous priorities in a Fiber Distributed Data Interface (FDDI) system. Based on an approximate analysis for the throughput under overload, two methods are derived. The first one is designed to meet requirements for the ratios between throughputs of stations with different priorities, whereas the second method includes absolute throughput values. Additionally, a proposition for dimensioning the timer thresholds of traffic classes that are deferred under overload is presented. Some examples for the application of these methods to overload control in FDDI systems are given. The results indicate that FDDI's timed token protocol provides the possibility to guarantee bandwidth not only for synchronous traffic, but also for asynchronous traffic. The allocation of guaranteed bandwidth can be controlled easily by selecting the timer thresholds according to the rules derived.>

The performance behavior of an FDDI (fiber distributed data interface) system with synchronous traffic as well as an extended FDDI (FDDI-II) system with isochronous and asynchronous traffic is evaluated. Multiple asynchronous priorities are considered, one priority level at each station. an iterative algorithm to calculate the first two moments of the cycle (i.e. token rotation) time is presented, and the major system characteristics are derived from the analysis in terms of queue-length distributions and mean waiting times of the packets. Since the model considers queues with limited buffer size, loss probabilities of data packets can also be calculated. Some results obtained by the approximate analysis are discussed using example configuration, and the accuracy is validated by detailed computer simulations. The complexity and convergence behavior of the algorithm is studied, and it is shown that the algorithm converges very quickly.>