Driven by recent advances in miniaturization and wireless communications, networks of small sensors can now work together to locate targets. This has promising applications in areas such as underwater localization, fault detection in industrial settings, and traffic surveillance. Even when no single sensor has enough information to determine a target’s position on its own, sensor networks can overcome this limitation by combining partial information from several vantage points.
In distributed networks, sensors must solve this problem cooperatively without relying on a centralized system. Complicating things further, the matching problem is somewhat circular; knowing which measurements belong to the same object requires knowing roughly where the objects are, but localizing the objects requires knowing how to combine the measurements. As more targets appear, the number of possible ways to match measurements grows exponentially. Existing distributed methods have mostly sidestepped this issue by either assuming the matching is already known, or by requiring highly capable sensors that can approximate target locations by themselves.
Against this backdrop, a research team led by Professor Weineng Chen from School of Computer Science and Engineering, South China University of Technology, China, has developed an innovative distributed optimization method called MASTER, which stands for ‘multi-agent swarm optimization method with contribution-based cooperation.’ The study was made available online on August 3, 2026, and was published in Volume 13, Issue 7 of IEEE/CAA Journal of Automatica Sinica on July 21, 2026.
The team’s key innovation was reformulating the localization task as a bilevel optimization problem, separating the search for target positions from measurement association among sensors. The association problem, which constitutes the ‘lower level’ of the optimization problem, can be solved deterministically using the well-established Kuhn–Munkres assignment algorithm. Instead of negotiating both target positions and a huge collection of possible measurement associations, sensors essentially negotiate only the candidate target positions, while the optimal local association is calculated separately.
The MASTER method comes into play at the ‘upper level’ of the optimization problem. In the proposed scheme, every sensor maintains an internal particle swarm of candidate position guesses, which it refines by alternating between local optimization and cooperation with neighboring sensors. During cooperation, sensors give more weight to information that recently contributed to significantly improving their local solutions. The method also adjusts how frequently sensors communicate; local search is emphasized early in the process, whereas cooperation is emphasized later to help the network reach a consensus.
In experiments involving different numbers of targets and sensors in both two- and three-dimensional settings, MASTER generally produced smaller localization errors than five existing distributed optimization methods. Its advantages became more pronounced as more sensors participated, suggesting that the method can make effective use of additional information from cooperating nodes. The adaptive communication strategy also helped sensors reach consensus more efficiently than a version using a fixed communication interval.
Notably, tests using round-trip propagation time, time-difference-of-arrival, received signal strength, and angle-of-arrival measurements showed that the proposed approach can work with several types of sensing data, highlighting its versatility. “Our algorithm could be easily integrated into existing systems because it does not rely on data-association information, central nodes, or highly capable devices,” remarks Prof. Chen.
Distributed optimization techniques will contribute to a more widespread adoption of sensor networks in industrial, surveillance, and environmental monitoring applications, especially in settings where a centralized implementation is impractical. The research team is already considering possible improvements to the proposed approach. “Future work could study how to divide the problem into multiple subproblems and conquer them cooperatively, thus improving the upper limit of localization capability. Further research will also consider more real-world factors, such as false alarms and tracking security,” concludes Prof. Chen.
***
Reference
Title of original paper: Multi-Agent Swarm Optimization Method with Contribution-Based Cooperation for Distributed Multi-Target Localization and Data Association
Journal: IEEE/CAA Journal of Automatica Sinica
About South China University of Technology, China
Tracing its roots back to 1918 and formally established in 1952 as one of China’s famed “Four Institutes of Technology,” South China University of Technology (SCUT) is a premier national public university directly under the Ministry of Education. A key member of China’s Double First-Class initiatives, SCUT is widely recognized as a flagship institute of technology in Southern China, known for its engineering excellence and integrated development across science, medicine, management, and the humanities. Today, SCUT educates over 52,000 full-time students across three campuses in Guangzhou. Backed by a strong faculty of over 4,400 staff, SCUT maintains robust global ties through academic partnerships with over 50 leading overseas institutions, cultivating talent and research within the dynamic Guangdong–Hong Kong–Macao Greater Bay Area.
Website: https://www.scut.edu.cn/en/
About Professor Weineng Chen from South China University of Technology , China
Dr. Weineng Chen received bachelor’s and Ph.D. degrees in computer science from Sun Yat-sen University in 2006 and 2012, respectively. Since 2016, he has been a Full Professor with the School of Computer Science and Engineering at South China University of Technology. He has co-authored over 200 international journal and conference papers, including more than 70 papers published in the IEEE Transactions journals. His current research interests include computational intelligence, swarm intelligence, network science, and their applications. He is currently the Vice-Chair of the IEEE Guangzhou Section, and the Chair of IEEE SMC Society Guangzhou Chapter.
About IEEE/CAA Journal of Automatica Sinica
IEEE/CAA Journal of Automatica Sinica is a journal of IEEE and CAA that publishes high-quality papers in English on original theoretical/experimental research and development in all areas of automation, including automatic control, artificial intelligence and intelligent control, systems theory and engineering, pattern recognition and intelligent systems, automation engineering and applications, information processing and information systems, network based automation, robotics, computer-aided technologies for automation systems, sensing and measurement, navigation, guidance, and control, smart city, smart grid, big data and data mining, Internet of Things, cyber-physical systems, blockchain, cloud computing for automation, and mechatronics.
Website: https://www.ieee-jas.net/indexen.htm
Funding information
This work was supported in part by the National Natural Science Foundation of China (62376097) and Guangdong Regional Joint Foundation Key Program (2022B1515120076).
IEEE/CAA Journal of Automatica Sinica
Experimental study
Not applicable
Multi-Agent Swarm Optimization Method with Contribution-Based Cooperation for Distributed Multi-Target Localization and Data Association
3-Aug-2026