Abstract. Networked control systems enable for flexible systems operation and reduce cost of installation and maintenance, potentially at the price of increasing the uncertainty due to information exchange over the network. We focus on the problem of information loss in terms of packet drops, which are modelled as stochastic events that depend on the current state of the network. To design reliable control systems the state of the network must be estimated online, together with the state of the controlled process. This paper proposes various approaches to discrete and hybrid stochastic estimation of network and process states, where the network is modelled as a Markov chain and the packet drop probability depends on the states of the Markov chain. The proposed techniques are evaluated on simulations and experimental data.

No abstract

No abstract