|
Comsim uses a complex radio propagation model, called knife-edge, to determine the probability of two sites communicating with each other. Variables include receiver and transmitter characteristics, power, and terrain. The simulation computes the Path Loss Capability (PLC) and compares it to the total attenuation over the path between the transmitter and the receiver. PLC must be greater than total attenuation for successful communication. Each variable used to compute PLC and total attenuation can be changed in an attempt to optimize their values for some particular cost-benefit function.
 The effects of terrain elevation and ground cover are taken into account. Communications ranges and sensor coverage are highly dependent upon terrain characteristics. To create the proper model, a seamless world database was assembled from many sources. Elements of this database include elevation, ground cover, and other features. Data sources include various formats from USGS , NIMA and other international sources. All are converted into a portable compressed internal format and stored for rapid access.
 The world database uses bilinear interpolations to increase elevation data resolution when the appropriate resolution is not available. Integration of geographic data into the simulation is eased because the data is handled by the same libraries that created the world database. The world database is stored in a machine independent fashion so it can be moved between processor types quickly and easily. All terrain databases use UTM coordinates to enable rapid line-of-sight calculations while avoiding spherical geometry problems inherent in geographic coordinate systems. Foliage and other ground cover is also accounted for.
 A typical sensing network has three levels with increasing communications power and bandwidth. Each level behaves as a gateway network to the less powerful level below it. The lowest level on the network are the least expensive type of radio, and are used solely for sensing.
There are several different ways to place a radio network over a piece of terrain. Each of these methods is selectable by the user as a laydown pattern. Optimal automated placement of a specified radio pattern is achieved through genetic search. Distributed computing is used to further speed the search process.
|
