A method for autonomous suppression of deception jamming based on large model collaborative reasoning

CN122339619APending Publication Date: 2026-07-03程湘儿

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Applications(China)
Current Assignee / Owner: 程湘儿
Filing Date: 2026-03-26
Publication Date: 2026-07-03

AI Technical Summary

Technical Problem

Existing deception interference suppression technologies are ill-suited to new and unknown interference, lack generalization ability, cannot achieve end-to-end autonomous closed loop, and lack multi-model collaborative reasoning mechanisms, resulting in poor identification and suppression effects.

Method used

A method based on large-scale collaborative reasoning is adopted to construct a hierarchical collaborative reasoning architecture of a central scheduling large model and multiple domain expert large models. Through collaborative reasoning in stages such as semantic encoding, interference identification, channel estimation, suppression parameter optimization and signal reconstruction, the full-link information sharing and closed-loop iteration are realized. Zero-shot learning and transfer learning are used to improve the recognition capability, and the model is autonomously iterated through a federated incremental learning framework.

Benefits of technology

It achieves accurate identification and efficient suppression of unknown new deception interference, adapts to complex interference scenarios, improves identification accuracy and suppression effect, and maintains continuous improvement in anti-interference capability in dynamically changing scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure CN122339619A_ABST

Patent Text Reader

Abstract

This invention belongs to the field of electronic countermeasures technology and discloses an autonomous suppression method for deception interference based on large-scale model collaborative reasoning. The central large-scale model completes global scheduling and initialization, and a high-speed cache shared memory pool is built to achieve full-link information sharing. Expert models such as interference identification and channel estimation perform parallel collaborative reasoning, adapting to different task requirements such as interference identification, parameter optimization, and signal reconstruction, ensuring the reasoning accuracy of each link. Through a unified semantic token sequence and interaction standard, the efficiency of multi-task collaboration is improved. First, multi-domain heterogeneous physical features are converted into a unified semantic sequence, and then the interference identification and channel estimation models interact bidirectionally and reason in parallel. This not only enables fine-grained analysis of the full-dimensional information such as the type, parameters, and attack intent of deception interference, but also completes accurate semantic annotation of unknown new interference. At the same time, confidence evaluation is added to the analysis results, improving the ability to identify dynamic composite modulation and zero-sample new deception interference.

Need to check novelty before this filing date? Find Prior Art