Tasking Distributed Coalition Sensor and Processing Assets to Perform Distributed Analytics Using a Vector Symbolic Architecture

Abstract In previous demonstrations we have shown how service workflows can be executed in a decentralized manner using a Vector Symbolic Architectures (VSA) to describe workflows as compact vector representations that enables dynamic service discovery and workflow execution in tactical edge wireless networks. In this demonstration we show how the VSA approach can be used to perform event detection by multicasting workflow vectors that task distributed coalition self-describing sensor and data processing services to self-organize into appropriate workflows to perform the specific event detection and processing tasks required. Previous work used internet protocol (IP) addressing to connect services together, however in this demonstration we show data can be routed between services using a truncated version of the service description vectors. We show how this approach is resilient to loss of services during workflow execution, either as a result of node failure or network fragmentation in contested environments. This is achieved through the automatic discovery of alternative services that can perform the same task and re-routing of data to these services. The demonstration uses components of the DAIS experimental framework, specifically a representative coalition edge network based on the ‘Anglova Scenario’ and implemented using the EMANE network emulator.
  • Chris Simpkin (Cardiff)
  • Ian Taylor (Cardiff)
  • Harrison Taylor (Cardiff)
  • Alun Preece (Cardiff)
  • Graham Bent (IBM UK)
  • Declan Millar (IBM UK)
  • Dave Conway-Jones (IBM UK)
  • Andreas Martens (IBM UK)
Date Sep-2020
Venue 4th Annual Fall Meeting of the DAIS ITA, 2020