Path loss predictions and measurements over urban and rural terrain at frequencies between 900 MHz and 28 GHz

摘要

We are concerned with the problem of predicting signal strengths to support rapid deployment of an outdoor wide area network in the aftermath of an attack or a natural disaster. Given that the user has only a digital elevation model (DEM) representation of the area, we ask how map detail, propagation model sophistication, and operating frequency are related to prediction accuracy. We are particularly concerned with signal loss on real paths obstructed by terrain, buildings, and vegetation rather than idealized line-of-sight (LOS) paths. Our assessment is based on a series of 257 signal strength measurements at frequencies of 902 MHz and 2.4, 24.1, and 27.5 GHz made for 63 locations on the Virginia Tech campus While many sophisticated techniques exist for calculating path loss and signal strength on shadowed (diffracted, obstructed) paths, often these are not of much practical use for the problem considered here because they are too general for cases in which the terrain is defined by the DEM; building heights may or may not be available, and there is little or no information about vegetation. The outlines of building footprints and terrain features are "fuzzy" limited in accuracy by the DEM resolution. To build a three-dimensional representation of an obstacle, the user must interpolate between the DEM values. The resulting representations rarely look like the knife-edges, wedges, and blocks common in theoretical analyses and useful for ray tracing in indoor environments. (We would expect the importance of the uncertainty in the size and shape of path obstacles to depend on frequency.) For all of these reasons, the user of an unnecessarily sophisticated model may invest a great deal of computation time to gain no significant increase in prediction accuracy. It is precisely that assertion that we have investigated in this paper.