REAL-TIME SPATIO-TEMPORAL DEDUPLICATION IN DISTRIBUTED MULTITARGET TRACKING VIA A BOUNDED-DEGRADATION ARCHITECTURE

Abstract

Distributed multi-target tracking systems tend to fall apart under heavy load: processing delays grow, tracks merge
into each other, and the root cause is usually unbounded algorithmic complexity combined with unpredictable memory
access patterns. This paper focuses on state estimation and spatio-temporal track deduplication across decentralized
processing nodes. The goal is to keep frame execution time deterministic and prevent identity swaps between tracks, even
during dense swarm engagements. To accomplish this, the study uses maneuvering target models reaching 10g
acceleration and simulated telemetry streams pushed to an 11× clutter saturation limit. The approach combines lock-free
ring buffering, structure-of-arrays memory layouts, flat prime-product spatial hash-grids with linear probing, and
localized joint probabilistic data association. The results show that a sub-10 ms processing budget holds at 5,500 clutter
points per frame, with a peak projected latency of 4.6 ms when all eight crossing locks are active simultaneously. Across
1,000 simulated trials of symmetric high-g crossings, the framework produced zero label swaps and maintained consistent
tracking through packet loss rates of up to 30%. The core scientific contribution is the formalization of the cache-hot
paradox: under heavy clutter, a bounded full-rank Cholesky factorization running entirely in L1 cache outperforms the
nominally cheaper Sherman-Morrison-Woodbury low-rank approximation, which gets bottlenecked by global memory
latency. A second contribution is an additive correction term for the NEES selection rule that absorbs the innovation
inflation caused by low-rank process noise approximations, preventing false ownership handoffs across lossy network
links.
Keywords: real-time target tracking, hardware-software co-design, distributed state estimation, cache-hot
paradox, joint probabilistic data association, track deduplication, bounded degradation, decentralized sensor networks.

References
1. Axis AB, "Multipath classification in radar detections," European Patent Application EP4092447A1, Nov. 23,
2022. Patent Identifier: EP4092447A1. https: // worldwide.espacenet.com / publicationDetails / biblio?II=0&ND=3&
adjacent=true&locale=en_EP&FT=D&date=20221123&CC=EP&NR=4092447A1&KC=A1.
2. Shenyang Ligong University, "Research on Fusion Tracking Technology in Heterogeneous Multi-Sensor,"
Academic Report, Oct. 2014. https://doi.org/10.4028/www.scientific.net/AMR.1056.158.
3. S. S. Blackman, "Multiple-Hypothesis Tracking for Multiple Target Tracking," IEEE Signal Processing
Magazine, vol. 21, no. 1, pp. 41-52, Jan. 2004. https://doi.org/10.1109/MAES.2004.1263228.
4. Y. Bar-Shalom, X. R. Li, and T. Kirubarajan, Estimation with Applications to Tracking and Navigation: Theory,
Algorithms and Software, Wiley-Interscience, 2001. https://doi.org/10.1002/0471221279.
5. X. R. Li and V. P. Jilkov, "Survey of Maneuvering Target Tracking. Part I: Dynamic Models," IEEE
Transactions on Aerospace and Electronic Systems, vol. 39, no. 4, pp. 1333-1364, Oct. 2003. https://doi.org/10.1109
/TAES.2003.1261132.
6. Fortem Technologies, Inc., "System and method for fusing asynchronous sensor tracks in a track fusion
application," U.S. Patent Application US20230215277A1, Feb. 20, 2023. Patent Identifier: US20230215277A1.
https://ppubs.uspto.gov/api/patents/html/US-20230215277-A1?source=US-PGPUB&requestToken=eyJzdWIiOiIwNjQyMjAyOC1hNDE5LTQ5NzgtYjk3Ni0zODhhNzY2ZGNkNzIiLCJ2ZXIiOiIwZDEzYTk0OC01M2NmLTRjYzYtYmU
xYy02OTA5YzJhNmU3NzIiLCJleHAiOjB9.
7. Dalian Maritime University, "Multi-Sensor Data Fusion Method Based on Improved Evidence Theory," Journal
of Marine Environmental Sensing, May 2023. https://doi.org/10.3390/jmse11061142.
8. A. Balachandran, A. Acharya, S. Chomal, and R. Tharmarasa, "Multipath-Assisted Multitarget Tracking with
Better Track Formation," IEEE Transactions on Aerospace and Electronic Systems, vol. 60, no. 3, pp. 3021-3037, June
2024. https://doi.org/10.1109/TAES.2024.3360012.
9. Politechnika Gdanska, "Sensors and Sensor’s Fusion in Autonomous Vehicles," Civil and Defense Navigation
Studies, Oct. 2021. https://doi.org/10.3390/s21196586.
10. Xian Electronic Engineering Research Institute, "Target tracking system for multi-sensor data fusion,"
Technical Report, Dec. 2017. https://doi.org/10.1109/ITNEC.2017.8285099.
11. Beihang University, "Study on Multi-Heterogeneous Sensor Data Fusion Method Based on Millimeter-Wave
Radar and Camera," Journal of Intelligent Transportation Systems, Jun. 2023. https://doi.org/10.3390/s23136044.
12. "An algorithm based on hierarchical clustering for multi-target tracking of multi-sensor data fusion," Journal
of Multi-Target Tracking, Jul. 2016. https://doi.org/10.1109/ChiCC.2016.7554147.
13. H. A. P. Blom and Y. Bar-Shalom, "The Interacting Multiple Model Algorithm for Systems with Markovian
Switching Coefficients," IEEE Transactions on Automatic Control, vol. 33, no. 8, pp. 780-783, Aug. 1988.
https://doi.org/10.1109/9.1299.
14. C. Hue, J.-P. Le Cadre, and P. Pérez, "Sequential Monte Carlo Methods for Multiple Target Tracking and Data
Association," IEEE Transactions on Signal Processing, vol. 50, no. 2, pp. 309-324, Feb. 2002. https://doi.org/10.1109/
78.978386.
15. Università degli Studi di Napoli Federico II, "Multi-sensor data fusion: A tool to enable UAS integration into
civil airspace," Journal of Aerospace Engineering, Oct. 2011. https://doi.org/10.1109/DASC.2011.6096082.
16. Chongqing Changan Automobile Co. Ltd., "Method and apparatus for associating radar target with visual
target, vehicle, and storage medium," Australian Patent AU2024200065A1, Jan. 04, 2024. Patent Identifier:
AU2024200065A. https://ipsearch.ipaustralia.gov.au/patents/2024200065.
17. Wuhan University of Technology, "Research on Target Detection Based on Distributed Track Fusion for
Intelligent Vehicles," Journal of Automated Driving Perception, Dec. 2019. https://doi.org/10.3390/s20010056.
18. Xi'an Jiaotong University, "Track-to-Track Association Based on Structural Similarity in the Presence of
Sensor Biases," Journal of Spatial and Coordinate Association, Jan. 2014. https://doi.org/10.1155/2014/294657.