Zhifan Song

Zhifan Song, Haralampos-G. Stratigopoulos, Hassan Aboushady

We present E-SpecFormer (Edge Spectrum monitoring Transformer) for end-to-end automatic modulation and covert channel (CC) recognition. We introduce LiTAN (Linear Tanh Attention Network), a Softmax- and LayerNorm-free attention mechanism that reduces complexity while increasing accuracy in RF tasks. E-SpecFormer is parameterized in four scalable variants (Nano, Small, Medium, Large) to accommodate diverse hardware constraints. Using the RadioML2018 dataset for modulation recognition, the Nano variant achieves 86.5% average accuracy for Signal-to-Noise Ratios (SNRs)>0 dB, and on the hardware Trojan (HT)-based CC dataset it reaches 94.2% accuracy, both with fewer than 10k parameters and up to speed of 92 μs per frame on FPGA/CPU co-execution, surpassing state-of-the-art edge models at a fraction of their cost. These results establish E-SpecFormer as an edge-efficient solution for real-time spectrum intelligence on Internet of Things (IoT) devices. GitHub link to the repository: https://github.com/zsniko/E-SpecFormer.

Xuhang Chen, Zhifan Song, Deyi Ji et al.

Large Language Models (LLMs) have exhibited impressive capabilities across diverse application domains. Recent work has explored Multi-LLM Agent Debate (MAD) as a way to enhance performance by enabling multiple LLMs to discuss and refine responses iteratively. Nevertheless, existing MAD methods predominantly focus on utilizing external structures, such as debate graphs, using LLM-as-a-Judge, while neglecting the application of self signals, such as token logits and attention, that arise during generation. This omission leads to redundant computation and potential performance degradation. In this paper, we shift the focus to the self signals of multi-LLM debate and introduce a Self-Signals Driven Multi-LLM Debate (SID), which leverages two types of self-signals: model-level confidence and token-level semantic focus, to adaptively guide the debate process. Our approach enables high-confidence agents to exit early at the model level and compress the redundant debate contents based on the attention mechanism. We evaluate our method on various LLMs and Multimodal LLMs across multiple challenging benchmarks. Experimental results demonstrate that our method not only outperforms existing MAD techniques in accuracy but also reduces token consumption, highlighting the effectiveness of utilizing self signals in enhancing both the performance and efficiency of multi-agent debate systems. Our code will be available at~\href{https://github.com/xuhang2019/SID}{\texttt{https://github.com/xuhang2019/SID}}.

Zhifan Song, Haralampos-G. Stratigopoulos, Hassan Aboushady

This paper presents a heterogeneous neural network accelerator for multi-task RF signal recognition, supporting automatic modulation recognition (AMR), hardware-Trojan covert channel (HT-CC) detection, and GNSS jamming classification. We introduce a compact attention-enhanced convolutional neural network (CNN) combined with LSDec, a learnable streaming decimator that enables adaptive temporal downsampling and flexible input lengths. The hardware architecture integrates a novel dual-pipeline, fused convolution-pooling engine with DMA-based streaming to minimize memory traffic and latency. Co-execution scheduling on the accelerator and SIMD-optimized CPU kernels reduces hardware resource usage while preserving high performance and task-level flexibility. Across three datasets, the proposed system achieves $\geq$ 99% average accuracy above 4 dB Signal-to-Noise Ratios (SNRs) on the RadioML2018 dataset for AMR, 90% on the HT-CC dataset, and 99.5% on the GNSS-Jamming dataset. The accelerator sustains an end-to-end inference latency of 98 $μ$s per frame, demonstrating its effectiveness for low-power, latency-critical multi-task spectrum-intelligence applications on embedded and edge devices.