Trust-worthy Coordinated Attack and Byzantine Agreement: Tools, Protocols, and Assertions

Trust-worthy distributed systems, such as the MDA Battle Manager (BM) System of Systems (SoS), require trusted coordination in the absence of ideal communication, i.e., when immediate and guaranteed broadcasting is not available. A fundamental issue for trust worthy distributed SoS is therefore attaining trustworthy agreement under real-life circumstances, when ideal communication does not exist. By agreement it is meant any agreement the BM might need to reach, such as agreement to particular weapon system to launch a kill. Two fundamental problems are: (i) Coordinated Attack (CA) -- when communication is unreliable and non-instantaneous how can a distributed team of generals decide on an attack, and (ii) Byzantine Agreement (BA) -- how can does this change if some generals are traitors (i.e., "faulty"). This proposal proposes working protocols and support tools for trust worthy CA and BA. Specifically, we suggest Phase I investigation and Phase II construction of: 1. Protocols using communicating UML statecharts augmented with recovery assertions. 2. Construction of a stochastic analysis simulator for the protocols of #1. The simulator will use item #3 below. 3. The extension of the StateRover white box test generator to generate a sample space for the stochastic analysis of protocols of item #1.