We modified the Transport and Application layers of GloMoSim to simulate a hypothetical sensor network environment consisting of nodes communicating over a 220 meter radius, which is typical for sensor nodes. The lower level protocols of our wireless network include Geographic Forwarding [16] over IP in the Network layer and 802.11B in the MAC layer. We simulated 32 byte packets sent in a 200 kbps wireless medium, a slightly larger bandwidth than the capabilities of today's sensor devices (50kbps-100kbps) to account for future improvements. At the physical layer, we use a Two-Ray Pathloss model with SNR-Bounded Noise at the receiving node. The model allows noise, attenuation, and subsequent loss on the wireless channel to be simulated, as opposed to perfect reception within a hypothetical radius.
Sensor nodes are equipped with sensors that poll their environment for specific events (e.g., acoustic sensors that monitor and can recognize certain acoustic signatures such as tank movement). In our experiments, a number of nodes are uniformly distributed in a 1,400 x 1,400 meter field. To test the ability of our architecture to track a moving target, we simulate an object moving across the field in a straight line. The moving object is tracked (presumably using acoustic or magnetic sensors) with a sensor polling period of 0.05 seconds, a granularity high enough to ensure up to date readings. Sensors register a target up to a sensing radius of approximately 100 meters.
For our tests we employ a simple application that computes an event's position through entity member reports to the entity leader. Entity members poll their sensors and send periodic updates to the corresponding entity leader notifying this leader of their current sensor reading and position. The leader computes the weighted average of the position of reporting entity members. The weighting is by sensor reading since higher readings presumably mean a closer target. This average value is sent back every 0.5 seconds to a ``friendly-force'' entity at a static location in the network. Upon receiving a report, this entity responds with a confirmation message sent back to the reporting entity. This feature allows us to test our architecture's ability to maintain end-to-end connectivity and forwarding between entities.