Organised by
In Association with
Supported by

SS18 - Advances in Distributed Kalman Filtering

The  rapid advances in sensor and communication technologies are   accompanied by an increasing demand for distributed state estimation   methods.  Centralized implementations of Kalman filter algorithms are  often too  costly in terms of communication bandwidth or simply  inapplicable - for  instance when mobile ad-hoc networks of autonomously  operating state  estimation systems are considered. Compared to  centralized approaches,  distributed or decentralized Kalman filtering  is considerably more  elaborate. In particular, the treatment of  dependent information shared  by different state estimation systems is a  central issue.   

Distributed  state estimation is, in general, a balancing act  between estimation  quality and flexible network design.  With the Distributed Kalman  Filter, it has been demonstrated that an  optimal (MSE minimal) estimate  can be computed in a distributed fashion,  but this algorithm is not  robust to packet delay and drops, node  failures, and changing network  topologies. However, in practice, these  problems deserve careful  attention and have to be addressed by future  research.  

Topics of interest

  • distributed and decentralized Kalman filters
  • track-to-track fusion 


distributed  Kalman filtering, common information, information  filtering,  track-to-track fusion, parallel Kalman filters, federated  Kalman  filtering, multisensor state estimation, fusion architectures,  channel  filtering, consensus Kalman filtering 

Special Session Organizers

  • Benjamin Noack, Karlsruhe  Institute of Technology (Germany)
  • Marc Reinhardt, Karlsruhe  Institute of Technology (Germany)
  • Uwe D. Hanebeck, Karlsruhe  Institute of Technology (Germany)
  • Felix Govaers, Fraunhofer Institute FKIE (Germany)
  • Alexander Charlish, Karlsruhe  Institute of Technology (Germany)

Special Session Contact