An incentive-based approach to distributed estimation with strategic sensors

Consider a setup in which a central estimator seeks to estimate a random variable using measurements from multiple sensors. The sensors incur an effort cost through consumption of resources to obtain a measurement. The sensors are self-interested and need to be compensated to generate measurements with a low enough error covariance that allows the calculation of an estimate with sufficient accuracy. However, a simple compensation scheme based on self-reported effort taken will not be sufficient, since both the quality of measurements taken by a sensor and the measurement values are private information for the sensor. A strategic sensor can misreport these values to increase his compensation. We formulate this problem as a contract design problem between the sensors and the central estimator and present an optimal contract between the central estimator and sensors as the solution.