Project number: MT-16-13

Funding source: Caltrans

Contract number: 65A0533

Funding amount: $87,703

Performance period: 3/2/2017 to 6/30/2018

Project description

For the last decade, advances in machine-learning algorithms provided easy and efficient ways to analyze large sets of data in search of correlations that would otherwise be extremely timeconsuming without the use of computers. The use of machine-learning algorithms for smart roads to track and analyze traffic attributes allows for highly accurate classifications while still being scalable and flexible enough to identify new types of vehicles that have yet to hit the market. Furthermore, extremely low power microprocessors have made it possible in the last few years to develop embedded systems that can run solely on battery power in multi-year applications without the necessity of recharging. Combined with low-power focused communications protocols and efficient vehicle classification algorithms, the lifetime of embedded systems can operate beyond a decade without any physical interaction after initial setup. This proposed research centers on the development of a distributed wireless sensing network that utilizes low power processors in conjunction to “in-sensor-node” machine learning algorithms for computation and a power-aware communications protocol for the development of a lightweight low-power multi-node MEMS sensing network. The collected data can be used for developing advanced models of urban traffic flow and for developing better policies to manage the impacts of transportation in metropolitan areas.