The proliferation of battlefield sensors and tactical data analytics enables the execution of various warfighting functions (such as target detection, identification, and localization) at an unprecedented speed and scale. A key challenge in realizing this potential is taming the combined latency of the various machine learning (ML) and signal processing algorithms, whose sum total will otherwise contribute significantly to the lag from data to decision-making. IoBT research significantly reduces the latencies involved in the sensor-to-decision loop by designing for timeliness and efficiency of intelligent processing on edge devices; rapidly creating IoBT networks capable of executing a variety of mission functions; reducing the amount of data required to accurately detect events of interest, and minimizing the data that need to be communicated for successful downstream inference.