用于自动调制识别的带关系令牌和残差分类器接口的统一主干-专家框架
A Unified Backbone--Expert Framework with Relation-Token and Residual--Classifier Interfaces for Automatic Modulation Recognition
浏览论文内容
中文总结 AI 辅助
针对自动调制识别的观测长度瓶颈，提出带关系令牌和残差分类器接口的统一主干-专家框架，在两个公开数据集上取得较高识别准确率，验证了专家-接口解耦架构的有效性。
中文摘要 AI 辅助
自动调制识别（AMR）在不同观测长度下面临独特的表示瓶颈，单一模型架构往往难以同时表现出色。为解决该问题，本文提出一种统一主干-专家框架，包含通用卷积状态空间主干与两个专用接口：针对短序列，在编码前注入显式的感知时延复平面描述符作为关系令牌，以补偿信息损失；针对长序列，设计门控多尺度残差细化模块修正特征图，结合固定平均分类器协作利用互补证据。该框架在RML2016.10b数据集上的总体平均准确率为67.28±0.14%，在HisarMod2019数据集上为87.19±0.77%（均值±三次运行的样本标准差）。通过三种子集消融实验、原生长度跨配置测试及受控窗口研究，进一步验证了该框架的有效性，确认专家-接口解耦相比一刀切架构的优势。
英文摘要
Automatic modulation recognition (AMR) faces distinct representation bottlenecks under varying observation lengths, where a single model architecture often fails to excel. To address this, we propose a unified backbone-expert framework with a common convolutional state-space backbone and two specialized interfaces. For short sequences, we inject explicit lag-aware complex-plane descriptors as relation tokens before encoding to compensate for information loss. For long sequences, we design a gated multi-scale residual refinement module to correct the feature map, combined with a fixed-averaging classifier collaboration to harness complementary evidence. Our framework achieves overall average accuracies of 67.28 \pm 0.14% on RML2016.10b and 87.19 \pm 0.77% on HisarMod2019 (mean \pm sample standard deviation over three runs), respectively. The framework's efficacy is further validated through three-seed ablations, native-length cross-configuration tests, and controlled window studies, confirming the benefit of expert-interface decoupling over one-size-fits-all architectures.
发表机构
- School of Mathematical Sciences, University of Electronic Science and Technology of China（电子科技大学数学科学学院）
- School of Information and Communication Engineering, University of Electronic Science and Technology of China（电子科技大学信息与通信工程学院）
机构由 AI 辅助整理，请以论文原文为准。
