CONSTRAINED SELECTIVE COMMUNICATION IN REAL-TIME INFORMATION-SYSTEMS

When information processing capability is limited, full and complete s haring of information is impossible. At issue is then which of the ava ilable messages should be communicated (and which sacrificed) and what the impact of different communication strategies may be - the so-call ed 'constrained selective communication' problem. This paper presents an analytical approach to the problem based on Shannon information the ory. Specifically, the paper examines three simple cases and reveals t heir analytical features, which are intuitive and supportive to some c ommon practice, such as the exception-reporting strategy. These three cases, in order of complexity, are: (1) the simple situation in which messages relate to the states of a single activity; (2) the extension of the first case to multiple activities, where these activities are a ssumed statistically independent; and (3) the situation with unequal c ommunication effort being required for different messages. Analytical results are obtained for the first case, and to some extent for the se cond case as well. To obtain optimal communication schemes for general cases, a solution method based on Lagrange relaxation is developed. A numerical example is provided to illustrate various issues this appro ach can address.