| Hi-index | 0.00 |
Extremely small, very low cost and very low power yet smart tactical sensors are commonly proposed. We have investigated the design patterns and algorithms that are capable of on-sensor dimensionality reduction so forwarding information rather than the raw data across these tactical sensor networks where data flows are hampered by a disconnected, intermittent and low bandwidth (DIL) operational domain. This work is novel because it has focused on what can be achieved with very simple sensor nodes that are scattered pseudo-randomly in a geographical region where each node is un-aware of its relative location with respect to a common reference point. We have investigated minimalist connectivity-based self-localization of nodes that can be used with extremely low hardware complexity and sensor insensitivity with the only constraint being that the sensor range is less than the communication connectivity range. These algorithms are less sensitive to the types of hybrid topology of sensors but more sensitive to the diversity of sensors. We have investigated extremely small footprint signal processing algorithms that are capable of maximizing knowledge extraction at the point of collection, which is wherever the sensor is deployed that can associate a signal to a specific source so that, if relevant, the information can be forwarded by the processing node so minimizing bandwidth.