Sensor networks and future applications

Pixellated detectors using integrated circuit (IC) electronics are becoming ubiquitous in particle physics, not only for tracking, but also in noble liquid readout planes, fast timing detectors, and high granularity calorimeters. The interconnections of such sensor arrays can form a network, which has traditionally been hard-wired and static. New development leverages communications network technology to add functionality not present in the individual sensors. This talk covers recent work and future directions in sensor networks for particle physics.

The DUNE near detector has implemented a network called Hydra for fault tolerant access to thousands of LArPix ICs in a cryogenic environment. Each IC has configurable access to its 4 neighbors, which allows to sidestep failed ICs, but the configuration must be loaded from outside to static memory in each IC. Ongoing development to make the configuration dynamically determined by the network itself will be presented. In addition to isolating failed ICs, such network intelligence can mitigate soft faults and balance data loads.

The nodes in a Hydra network are ICs, not individual sensor channels. However, in particle tracking pixel detectors, a single IC has thousands of pixels with high internal data rate. The development on network-on-chip (NOC) is being pursued to address problems faced by pixel trackers, such as radiation-induced single event effects (SEE). An inherently fault tolerant mesh network architecture could save years of design work to discover and patch SEE weak points for future rad-hard ICs and eliminate potential hidden failure mechanisms, in addition to providing load balancing for optimal bandwidth utilization. A further benefit would mitigate the impact of manufacturing defects, which currently precludes the use of wafer-scale sensors in large systems.

The above networks are all 2-dimensional grid arrangements with 4 neighbors per node. A more general case, with arbitrary connectivity, will be discussed, which would be natural for proposed novel sensors such as smart dust, consisting of tiny single pixel ICs with wireless power and connectivity.