WiSE - A Wireless System Engineering Tool

1. Overview

Imagine a large department store (Walmart? Bloomingdales?) that wants to equip sales people with portable point of sale terminals so they can serve customers anywhere in the building. Imagine a supermarket with shelf tags that can receive signals from and send signals to base stations in the ceiling. Imagine a convention center or a large office building with continually changing network and voice communication needs. In each case, a system that uses indoor radio-frequency transmitters and receivers can be a cost-effective way to provide mobility and easy rearrangement, far more flexibly than any wired system.

In general, the approach is to position base stations carefully so that they cover the building adequately. Thus the design problem is: given a building, where do we position how many base stations of what power to cover the building to a specified minimum power level at a minimum cost?

The problem is far too complicated for any simple approximations; modern buildings use many materials, which degrade performance by attenuation, reflections, multipath, and the like. It is much too time-consuming and expensive to perform experiments on the premises of potential customers; even the taking of measurements is costly and slow. Brute force solutions with high-power transmitters are infeasible, for reasons of cost and public concern over electromagnetic radiation; in any case, the field is too competitive to permit over-engineered solutions. We need a tool that can provide a very good design quickly, with minimal human interaction. It should design small systems to work almost ``out of the box,'' while large systems should require only limited tuning before a sound proposal can be written.

Wireless System Engineering (WiSE) is a system that attacks this combination of problems; it has five major components.

"Propagation Model" : A propagation model predicts the local mean of power received at any given point. For each point, we compute the scalar sum of the powers of multipath components reaching the specified location. The model includes the effects of angle of incidence, polarization, material dielectric constant, and antenna patterns. Predictions can also account for diffraction effects around corners, which is particularly significant for outdoor microcell environments.

"Physical Database Acquisition" : The building database, i.e., the locations and radio characteristics of all walls, floors and ceilings, is acquired and digitized. Paper or electronic floorplans are entered into a Computer Aided Design (CAD) environment and relevant building information is extracted.

"Propagation Prediction" : The propagation model is essentially a 3-d ray-tracing process that in principle follows all paths from a base station to a portable. A brute-force approach here requires far too much computation time to be feasible for realistic buildings. Accordingly, a new algorithm has been developed to reduce the number of path components that have to be evaluated to a manageable level; this typically reduces computation time by two orders of magnitude. A complete coverage map of a building with several hundred walls can now be computed in less than a minute on a standard workstation.

"Port Placement Optimization" : An optimization algorithm determines, for given choices of parameters like transmitter frequency and power, receiver sensitivity, signal/noise ratios, etc., a near-optimum placement of a number of base stations to satisfy requirements on minimum signal strength at each potential portable location.

"Graphical User Interface" : Finally, a graphical user interface provides convenient access to these modeling tools. The interface displays plan, elevation and perspective views of a building, upon which it will superimpose all rays among any set of base stations and portables, or a complete coverage map with a color scale indicating signal strength. Views may be scaled and panned over. Parameters like wavelength, power levels, antenna types, and so on may be set from popup menus or dialog boxes. Base stations and portable units may be positioned interactively; the results of previous computations may be displayed for comparison. The progress of prediction or optimization may be monitored as it proceeds asynchronously.

The following sections describe these components in more detail, and report on computational experience.

Previous Beginning Next