Goals of Sensor Networks

Geographical mapping: Exploration and mapping of geographical information, including topological information, surface conditions, density, etc.
Distributed sample/data collection: Collecting soil samples, images, atmospheric measurements, etc.
Finding target object types: Detecting water, minerals, geothermal activity, life, etc.
Performing experiments: Performing on-site experiments at locations of interest.

Types of Sensor Agents

Oribiting station: Satellite orbiting station to accept commands from Earth and relay data back to Earth, for example.
Base station(s): Ground stations responsible for accepting and relaying data to orbiting station.
Carriers: Long-range vehicles which carry smaller agents.
Mappers: Rovers responsible for mapping the geographical topography
Local Positioning System (LPS) transmitters: Ad-hoc local transmitters (arranged in a triangular configuration) used to construct accurate positional information for other agents.
Mobile lab rovers: Rovers which conduct experiments.
Power sources: Agents which generate/provide power (solar, geothermal, battery, etc.)

More than one of these functions may be combined in one physical agent.

Ad-hoc networking. Ability to relay information in an ad-hoc network, with dynamic upload stations and minimal centralized control.
Location awareness. Agents need to be aware of their positions relative to other agents, either through adjacent signal detection, or through a local positioning system (LPS).

Coordination of diverse elements. System must be able to coordinate agents with diverse functions (as listed in Types of Sensor Agents).

Staggered lifetimes. Different groups of agents may be activated at different times. One group's completion may trigger another group's activitation. Lifetimes of different groups may also be staggered to prolong mission life.

Predictive coordination. System must adaptively use algorithms to coordinate multiple agents to meet conditions on-hand. For example, multiple agents may self-coordinate to explore a cave by forming an arm of the ad-hoc network into the cave.
Environmental learning. System must collectively learn to avoid hazards, navigate the topography, zone in on interesting areas, etc.

Integrated remote/autonomous control. Since sensor networks may be deployed in locations not permitting real-time control, such systems must support automomous control guided by high-level remote control commands.
Safe bootstrapping. System must have capability to be reset to a safe operating state.
Dynamic reconfiguration. System must have capability to be dynamically reconfigured by downloading new software.

Minimal centralized control. Avoid centralized control of groups of agents to avoid a single point of failure.
Power management. Methods to conserve power, and to efficiently distribute power, may prove essential.
Fault recovery. The ability to recover from unseen obstacles, faults, accidents, etc, using techniques such as physical recovery, control reset, etc. Backup procedures when primary procedures fail are also essential.

Communications encryption. May want to encrypt communication if sensor network is deployed in enemy territory.
Timed expiration. Software or hardware needs to self-destruct to avoid being stolen by enemies.