California State University, Long Beach
Inside CSULB Logo

Smaller Sensors To Monitor Traffic More Efficiently

Published: April 4, 2016

Advancements in microelectro-mechanical systems and wireless communication have motivated the development of small and low-power sensors and radio-equipped modules which are now replacing traditional wired sensor systems. In the last decade, the landscape of wireless sensor network applications has been extending rapidly in many fields such as factory and building automation, environmental monitoring, security systems and in a wide variety of commercial and military areas.

Mohammad Mozumdar, a member of the university in the Department of Electrical Engineering since 2012, has been developing a micro-wireless sensing system that will be implanted underneath the pavement of roadways to detect and “classify” vehicles. He received a $35,000 DENSO grant to equip his lab for Smart Sensing Education and Training. Moreover, he had a research grant ($34,995) from METRANS in 2015 in collaboration with the Department of Transportation and Caltrans with the goal to design a smaller, more efficient way to monitor and classify freeway and surface road traffic.

Currently, many transportation agencies implant vehicle detection sensors underneath the road pavement to collect traffic data. However, these agencies are using inefficient, decades-old sensing systems, such as the inductive loop and similar technology. Due to its large physical area and bulk size, the current system technology is significantly expensive with transportation agencies spending millions of tax-payer dollars yearly to install and maintain this outdated sensing system, according to Mozumdar. The potential exists for an advanced system to be designed that could not only reduce costs dramatically but could also provide efficient real-time data. Motivated by a need to enhance and modernize the transportation system, Mozumdar and his research team have been working to develop an advance modular traffic system that would be a milestone achievement towards smart road and traffic engineering.

Mozumdar explained that the inductive loop lies beneath the freeway’s pavement where it creates a magnetic field.

“If metal passes over the loop, it creates a distortion in the magnetic field,” he said. “Based on that distortion, we know this indicates a car passing. The current design for an inductive loop is huge. Some are six feet by six feet and others as large as 10 feet by 10 feet.”

The newest design by Mozumdar’s research group is smaller, much smaller.

“What we’ve done in the Smart Sensing Education and Training Lab is to design a quarter-sized system to replace the current inductive loop,” he continued. “It will dramatically change the loop’s installation and begin a new era of technology in traffic detection. Not only does the design system created in the CSULB lab detect passing vehicles but it can tell what type of vehicle. It knows whether it has been passed by a sedan or a truck.”

Another important feature is that, right now, the installation of the inductive loop is “humongous,” he said.

“The loop is not only big but electricity must be provided,” said Mozumdar. “Our system is based on the battery. We are designing power-aware algorithms that enable the system to run on batteries year after year. With battery and wrapping shield, this designed system could be around four-by-four inches. Compare that to a six-foot by six-foot sensor today. Caltrans uses sensors on every street in the state and it costs millions of dollars to install and maintain them. If we can move forward with a more efficient product, it can save humongous amounts of money that could be spent for many other purposes.”

The key to the sensor is a machine learning algorithm.

“Our approach for vehicle classification utilizes J48 classification algorithm, an extension of decision tree machine learning ID3 approach,” he said. “The result of our experiment shows that the vehicle classification system is effective and efficient with the accuracy at nearly 100 percent.”

Working on the project are (l-r) Ankit Trivedi, Kuntal Patel, assistant professor Mohammad Mozudmar, Indraneel Bavkar, Nitish Bhardwaj and Amal Francis.
Working on the project are (l-r) Ankit Trivedi, Kuntal Patel, assistant professor Mohammad Mozudmar, Indraneel Bavkar, Nitish Bhardwaj and Amal Francis.

At first, the lab tested various types of vehicles in order to “train” the algorithm to understand from the magnetic distortion what kind of vehicle had just passed. The students later published a peer-reviewed paper about their sensor research in the 2015 IEEE Green Energy and Systems conference. Currently six students are conducting field tests on campus in cooperation with CSULB’s Parking and Transportation Services to modify their algorithm. Once the sensor is developed, statewide use could follow.

“I’ve been in contact with deputy director of public works for the City of Long Beach Sean Crumby who has promised to help our students install the new sensors on Bellflower Boulevard this spring,” said Mozudmar. “Tyler Reeb, the director of research for CSULB’s Center for International Trade and Transportation, is coordinating this effort and also providing valuable input to move this research from the lab to the real world.”

Mozumdar believes the potential for the new smaller sensors is great and wide reaching.

“Southern California plays host to two big ports—Long Beach and Los Angeles—and it is important to keep track of the heavy truck traffic,” he said. “This technology helps the policy maker to understand how many and what kinds of vehicles are running on which highway in a 24-hour time frame. Based on that, policy can be made.”

Mozumdar credits the technical sophistication of the Smart Sensing Education and Training lab for its sensor-designing success.

“You need a complete set of tools to create this level of miniaturization,” he explained, pointing to the lab’s high-end milling machine and a “pick-and-place machine” used to place surface-mount devices onto a printed circuit board. “These machines were purchased by multiple grants,” he said. “The pick-and-place machine alone cost $27,000 and the milling machine cost $22,000. With this sort of advanced technology, students are able to design the kinds of systems that can be used to start their own companies.”

One of main goals this research, Mozumdar said, is to lengthen the sensors’ longevity wanting to make sure the sensors can run year after year.

“It is the same principle as a laptop computer’s energy saver that switches off the screen after a certain amount of time,” he said. “We need to come up with the algorithms to prolong the sensor’s lifetime.”

Mozumdar also thanked College of Engineering Dean Forouzan Golshani and associate dean of Research Hamid Rahai for their generous support in this research project.