Reliable deployment of automatic vehicle identification sensors for origin-destination matrix observation

An Origin-Destination Matrix (ODM) is fundamental in transportation studies, as it provides essential insights into travel patterns and demand. The ODM can be constructed using data collected from Automatic Vehicle Identification (AVI) sensors strategically installed on network links. However, the quality of the ODM can be significantly affected by the fact that sensors are subject to failure in real-world scenarios. This issue underscores the importance of ODM reliability because it can significantly influence the study outcomes. Some researchers focused on incorporating sensor failure considerations into the Sensor Location Problem for ODM observation. One common approach is to consider a predefined level of reliability for ODM and try to find a sensor deployment with the minimum number of sensors that meet the reliability constraint. In this study, we first show by a counter-example that the most recently developed reliable sensor location model for ODM observation using the mentioned approach does not guarantee the predefined reliability level for ODM. Second, we introduce a new reliability term for ODM observation and incorporate it into our new reliable AVI sensor location models specifically designed to observe ODM and route flows. Additionally, we develop reliable AVI sensor location models that accommodate partial observations of ODM and route flows while adhering to budget constraints. Third, a greedy algorithm and a Genetic-Based Algorithm (GBA) are developed to solve the proposed models for middle to large-scale networks. Finally, the proposed models are applied to some numerical examples to illustrate their applicability and effectiveness. The numerical examples revealed the models’ capability to identify optimal sensor locations for reliable observation of ODM and route flows considering sensor failure. Moreover, the results highlighted the efficiency of the GBA as an efficient solution method, especially for medium and large-scale networks.