Reconfigurable intelligent surfaces (RIS) have emerged as a transformative technology for electromagnetic (EM) wave manipulation, offering unprecedented control over wave reflections compared to traditional metallic reflectors. By utilizing an array of tunable elements, RIS can steer and shape electromagnetic waves to enhance signal quality in wireless communication and radar systems. However, practical implementations face significant challenges due to hardware limitations and phase quantization errors. In this work, a 1-bit RIS prototype operating at 28 GHz is developed to experimentally evaluate the impact of hardware constraints on RIS performance. Unlike conventional studies that model RIS as an ideal phase-shift matrix, this study accounts for physical parameters that influence the actual reflection pattern. In particular, the presence of specular reflection due to hardware limitations is investigated. Additionally, the effects of phase quantization errors, which stem from the discrete nature of RIS elements, are analyzed, and a genetic algorithm (GA)-based optimization is introduced to mitigate these errors. The proposed optimization strategy effectively reduces gain degradation at the desired angle caused by 1-bit quantization, enhancing the overall performance of RIS. The effectiveness of the approach is validated through measurements, underscoring the importance of advanced phase control techniques in improving the functionality of RIS.
The rapid growth of data traffic and the emerging AI-native wireless architectures in NextG cellular systems place new demands on the fronthaul links of Cloud Radio Access Networks (C-RAN). In this paper, we investigate neural compression techniques for the Common Public Radio Interface (CPRI), aiming to reduce the fronthaul bandwidth while preserving signal quality. We introduce two deep learning-based compression algorithms designed to optimize the transformation of wireless signals into bit sequences for CPRI transmission. The first algorithm utilizes a non-linear transformation coupled with scalar/vector quantization based on a learned codebook. The second algorithm generates a latent vector transformed into a variable-length output bit sequence via arithmetic encoding, guided by the predicted probability distribution of each latent element. Novel techniques such as a shared weight model for storage-limited devices and a successive refinement model for managing multiple CPRI links with varying Quality of Service (QoS) are proposed. Extensive simulation results demonstrate notable Error Vector Magnitude (EVM) gains with improved rate-distortion performance for both algorithms compared to traditional methods. The proposed solutions are robust to variations in channel conditions, modulation formats, and noise levels, highlighting their potential for enabling efficient and scalable fronthaul in NextG AI-native networks as well as aligning with the current 3GPP research directions.
To leverage high-frequency bands in 6G wireless systems and beyond, employing massive multiple-input multipleoutput (MIMO) arrays at the transmitter and/or receiver side is crucial. To mitigate the power consumption and hardware complexity across massive frequency bands and antenna arrays, a sacrifice in the resolution of the data converters will be inevitable. In this paper, we consider a point-to-point massive MIMO system with 1-bit digital-to-analog converters at the transmitter, where the linearly precoded signal is supplemented with dithering before the 1-bit quantization. For this system, we propose a new maximumlikelihood (ML) data detection method at the receiver by deriving the mean and covariance matrix of the received signal, where symbol-dependent linear minimum mean squared error estimation is utilized to efficiently linearize the transmitted signal. Numerical results show that the proposed ML method can provide gains of more than two orders of magnitude in terms of symbol error rate over conventional data detection based on soft estimation.
Accurate channel estimation is critical for realizing the performance gains of massive multiple-input multiple-output (MIMO) systems. Traditional approaches to channel estimation typically assume ideal receiver hardware and linear signal models. However, practical receivers suffer from impairments such as nonlinearities in the low-noise amplifiers and quantization errors, which invalidate standard model assumptions and degrade the estimation accuracy. In this work, we propose a nonlinear channel estimation framework that models the distortion function arising from hardware impairments using Gaussian process (GP) regression while leveraging the inherent sparsity of massive MIMO channels. First, we form a GP-based surrogate of the distortion function, employing pseudo-inputs to reduce the computational complexity. Then, we integrate the GP-based surrogate of the distortion function into newly developed enhanced sparse Bayesian learning (SBL) methods, enabling distortion-aware sparse channel estimation. Specifically, we propose two nonlinear SBL methods based on distinct optimization objectives, each offering a different trade-off between estimation accuracy and computational complexity. Numerical results demonstrate significant gains over the Bussgang linear minimum mean squared error estimator and linear SBL, particularly under strong distortion and at high signal-to-noise ratio.
One challenge of integrating speech input with large language models (LLMs) stems from the discrepancy between the continuous nature of audio data and the discrete token-based paradigm of LLMs. To mitigate this gap, we propose a method for integrating vector quantization (VQ) into LLM-based automatic speech recognition (ASR). Using the LLM embedding table as the VQ codebook, the VQ module aligns the continuous representations from the audio encoder with the discrete LLM inputs, enabling the LLM to operate on a discretized audio representation that better reflects the linguistic structure. We further create a soft "discretization" of the audio representation by updating the codebook and performing a weighted sum over the codebook embeddings. Empirical results demonstrate that our proposed method significantly improves upon the LLM-based ASR baseline, particularly in out-of-domain conditions. This work highlights the potential of soft discretization as a modality bridge in LLM-based ASR.
The high computational costs associated with large deep learning models significantly hinder their practical deployment. Model pruning has been widely explored in deep learning literature to reduce their computational burden, but its application has been largely limited to computer vision tasks such as image classification and object detection. In this work, we propose a structured pruning method targeted for Learned Image Compression (LIC) models that aims to reduce the computational costs associated with image compression while maintaining the rate-distortion performance. We employ a Neural Architecture Search (NAS) method based on the rate-distortion loss for computing the pruning ratio for each layer of the network. We compare our pruned model with the uncompressed LIC Model with same network architecture and show that it can achieve model size reduction without any BD-Rate performance drop. We further show that our pruning method can be integrated with model quantization to achieve further model compression while maintaining similar BD-Rate performance. We have made the source code available at gitlab.com/viper-purdue/lic-pruning.
Despite its improvements in coding performance compared to traditional codecs, Learned Image Compression (LIC) suffers from large computational costs for storage and deployment. Model quantization offers an effective solution to reduce the computational complexity of LIC models. However, most existing works perform fixed-precision quantization which suffers from sub-optimal utilization of resources due to the varying sensitivity to quantization of different layers of a neural network. In this paper, we propose a Flexible Mixed Precision Quantization (FMPQ) method that assigns different bit-widths to different layers of the quantized network using the fractional change in rate-distortion loss as the bit-assignment criterion. We also introduce an adaptive search algorithm which reduces the time-complexity of searching for the desired distribution of quantization bit-widths given a fixed model size. Evaluation of our method shows improved BD-Rate performance under similar model size constraints compared to other works on quantization of LIC models. We have made the source code available at gitlab.com/viper-purdue/fmpq.
This paper considers the problem of decentralized optimization on compact submanifolds, where a finite sum of smooth (possibly non-convex) local functions is minimized by $n$ agents forming an undirected and connected graph. However, the efficiency of distributed optimization is often hindered by communication bottlenecks. To mitigate this, we propose the Quantized Riemannian Gradient Tracking (Q-RGT) algorithm, where agents update their local variables using quantized gradients. The introduction of quantization noise allows our algorithm to bypass the constraints of the accurate Riemannian projection operator (such as retraction), further improving iterative efficiency. To the best of our knowledge, this is the first algorithm to achieve an $\mathcal{O}(1/K)$ convergence rate in the presence of quantization, matching the convergence rate of methods without quantization. Additionally, we explicitly derive lower bounds on decentralized consensus associated with a function of quantization levels. Numerical experiments demonstrate that Q-RGT performs comparably to non-quantized methods while reducing communication bottlenecks and computational overhead.
We develop a communication-theoretic framework for privacy-aware and resilient decision making in cyber-physical systems under misaligned objectives between the encoder and the decoder. The encoder observes two correlated signals ($X$,$\theta$) and transmits a finite-rate message $Z$ to aid a legitimate controller (the decoder) in estimating $X+\theta$, while an eavesdropper intercepts $Z$ to infer the private parameter $\theta$. Unlike conventional setups where encoder and decoder share a common MSE objective, here the encoder minimizes a Lagrangian that balances legitimate control fidelity and the privacy leakage about $\theta$. In contrast, the decoder's goal is purely to minimize its own estimation error without regard for privacy. We analyze fully, partially, and non-revealing strategies that arise from this conflict, and characterize optimal linear encoders when the rate constraints are lifted. For finite-rate channels, we employ gradient-based methods to compute the optimal controllers. Numerical experiments illustrate how tuning the privacy parameter shapes the trade-off between control performance and resilience against unauthorized inferences.
Regardless of the advancements in device capabilities, efficient inferencing advanced large language models (LLMs) at the edge remains challenging due to limited device memory and power constraints. Existing strategies, such as aggressive quantization, pruning, or remote inference, trade accuracy for efficiency or lead to substantial cost burdens. This position paper introduces a new approach that leverages speculative decoding, previously viewed primarily as a decoding acceleration technique for autoregressive generation of LLMs, as a promising approach specifically adapted for edge computing by orchestrating computation across heterogeneous devices. We propose \acronym, a method that allows lightweight edge devices to draft multiple candidate tokens locally using diverse draft models, while a single, shared edge server efficiently batches and verifies the tokens utilizing a more precise target model. This approach supports device heterogeneity and reduces server-side memory footprint by avoiding the need to deploy multiple target models. Our initial experiments with Jetson Orin Nano, Raspberry Pi 4B/5, and an edge server equipped with 4 Nvidia A100 GPUs indicate substantial benefits: significantly increased system throughput, capacity, and better cost efficiency, all without sacrificing model accuracy.
This paper introduces EdgeProfiler, a fast profiling framework designed for evaluating lightweight Large Language Models (LLMs) on edge systems. While LLMs offer remarkable capabilities in natural language understanding and generation, their high computational, memory, and power requirements often confine them to cloud environments. EdgeProfiler addresses these challenges by providing a systematic methodology for assessing LLM performance in resource-constrained edge settings. The framework profiles compact LLMs, including TinyLLaMA, Gemma3.1B, Llama3.2-1B, and DeepSeek-r1-1.5B, using aggressive quantization techniques and strict memory constraints. Analytical modeling is used to estimate latency, FLOPs, and energy consumption. The profiling reveals that 4-bit quantization reduces model memory usage by approximately 60-70%, while maintaining accuracy within 2-5% of full-precision baselines. Inference speeds are observed to improve by 2-3x compared to FP16 baselines across various edge devices. Power modeling estimates a 35-50% reduction in energy consumption for INT4 configurations, enabling practical deployment on hardware such as Raspberry Pi 4/5 and Jetson Orin Nano Super. Our findings emphasize the importance of efficient profiling tailored to lightweight LLMs in edge environments, balancing accuracy, energy efficiency, and computational feasibility.
Diffusion models represent the cutting edge in image generation, but their high memory and computational demands hinder deployment on resource-constrained devices. Post-Training Quantization (PTQ) offers a promising solution by reducing the bitwidth of matrix operations. However, standard PTQ methods struggle with outliers, and achieving higher compression often requires transforming model weights and activations before quantization. In this work, we propose HadaNorm, a novel linear transformation that extends existing approaches and effectively mitigates outliers by normalizing activations feature channels before applying Hadamard transformations, enabling more aggressive activation quantization. We demonstrate that HadaNorm consistently reduces quantization error across the various components of transformer blocks, achieving superior efficiency-performance trade-offs when compared to state-of-the-art methods.
The Segment Anything Model 2 (SAM2) has gained significant attention as a foundational approach for promptable image and video segmentation. However, its expensive computational and memory consumption poses a severe challenge for its application in resource-constrained scenarios. In this paper, we propose an accurate low-bit quantization method for efficient SAM2, termed Q-SAM2. To address the performance degradation caused by the singularities in weight and activation distributions during quantization, Q-SAM2 introduces two novel technical contributions. We first introduce a linear layer calibration method for low-bit initialization of SAM2, which minimizes the Frobenius norm over a small image batch to reposition weight distributions for improved quantization. We then propose a Quantization-Aware Training (QAT) pipeline that applies clipping to suppress outliers and allows the network to adapt to quantization thresholds during training. Our comprehensive experiments demonstrate that Q-SAM2 allows for highly accurate inference while substantially improving efficiency. Both quantitative and visual results show that our Q-SAM2 surpasses existing state-of-the-art general quantization schemes, especially for ultra-low 2-bit quantization. While designed for quantization-aware training, our proposed calibration technique also proves effective in post-training quantization, achieving up to a 66% mIoU accuracy improvement over non-calibrated models.
Video matting is crucial for applications such as film production and virtual reality, yet deploying its computationally intensive models on resource-constrained devices presents challenges. Quantization is a key technique for model compression and acceleration. As an efficient approach, Post-Training Quantization (PTQ) is still in its nascent stages for video matting, facing significant hurdles in maintaining accuracy and temporal coherence. To address these challenges, this paper proposes a novel and general PTQ framework specifically designed for video matting models, marking, to the best of our knowledge, the first systematic attempt in this domain. Our contributions include: (1) A two-stage PTQ strategy that combines block-reconstruction-based optimization for fast, stable initial quantization and local dependency capture, followed by a global calibration of quantization parameters to minimize accuracy loss. (2) A Statistically-Driven Global Affine Calibration (GAC) method that enables the network to compensate for cumulative statistical distortions arising from factors such as neglected BN layer effects, even reducing the error of existing PTQ methods on video matting tasks up to 20%. (3) An Optical Flow Assistance (OFA) component that leverages temporal and semantic priors from frames to guide the PTQ process, enhancing the model's ability to distinguish moving foregrounds in complex scenes and ultimately achieving near full-precision performance even under ultra-low-bit quantization. Comprehensive quantitative and visual results show that our PTQ4VM achieves the state-of-the-art accuracy performance across different bit-widths compared to the existing quantization methods. We highlight that the 4-bit PTQ4VM even achieves performance close to the full-precision counterpart while enjoying 8x FLOP savings.
Deep neural network (DNN) quantization for fast, efficient inference has been an important tool in limiting the cost of machine learning (ML) model inference. Quantization-specific model development techniques such as regularization, quantization-aware training, and quantization-robustness penalties have served to greatly boost the accuracy and robustness of modern DNNs. However, very little exploration has been done on improving the initial conditions of DNN training for quantization. Just as random weight initialization has been shown to significantly impact test accuracy of floating point models, it would make sense that different weight initialization methods impact quantization robustness of trained models. We present an extensive study examining the effects of different weight initializations on a variety of CNN building blocks commonly used in efficient CNNs. This analysis reveals that even with varying CNN architectures, the choice of random weight initializer can significantly affect final quantization robustness. Next, we explore a new method for quantization-robust CNN initialization -- using Graph Hypernetworks (GHN) to predict parameters of quantized DNNs. Besides showing that GHN-predicted parameters are quantization-robust after regular float32 pretraining (of the GHN), we find that finetuning GHNs to predict parameters for quantized graphs (which we call GHN-QAT) can further improve quantized accuracy of CNNs. Notably, GHN-QAT shows significant accuracy improvements for even 4-bit quantization and better-than-random accuracy for 2-bits. To the best of our knowledge, this is the first in-depth study on quantization-aware DNN weight initialization. GHN-QAT offers a novel approach to quantized DNN model design. Future investigations, such as using GHN-QAT-initialized parameters for quantization-aware training, can further streamline the DNN quantization process.
Large language models (LLMs) have demonstrated exceptional performance across a variety of tasks. However, their substantial scale leads to significant computational resource consumption during inference, resulting in high costs. Consequently, edge device inference presents a promising solution. The primary challenges of edge inference include memory usage and inference speed. This paper introduces MNN-LLM, a framework specifically designed to accelerate the deployment of large language models on mobile devices. MNN-LLM addresses the runtime characteristics of LLMs through model quantization and DRAM-Flash hybrid storage, effectively reducing memory usage. It rearranges weights and inputs based on mobile CPU instruction sets and GPU characteristics while employing strategies such as multicore load balancing, mixed-precision floating-point operations, and geometric computations to enhance performance. Notably, MNN-LLM achieves up to a 8.6x speed increase compared to current mainstream LLM-specific frameworks.
Discrete audio tokens are compact representations that aim to preserve perceptual quality, phonetic content, and speaker characteristics while enabling efficient storage and inference, as well as competitive performance across diverse downstream tasks. They provide a practical alternative to continuous features, enabling the integration of speech and audio into modern large language models (LLMs). As interest in token-based audio processing grows, various tokenization methods have emerged, and several surveys have reviewed the latest progress in the field. However, existing studies often focus on specific domains or tasks and lack a unified comparison across various benchmarks. This paper presents a systematic review and benchmark of discrete audio tokenizers, covering three domains: speech, music, and general audio. We propose a taxonomy of tokenization approaches based on encoder-decoder, quantization techniques, training paradigm, streamability, and application domains. We evaluate tokenizers on multiple benchmarks for reconstruction, downstream performance, and acoustic language modeling, and analyze trade-offs through controlled ablation studies. Our findings highlight key limitations, practical considerations, and open challenges, providing insight and guidance for future research in this rapidly evolving area. For more information, including our main results and tokenizer database, please refer to our website: https://poonehmousavi.github.io/dates-website/.
To address the enormous size of Large Language Models (LLMs), model compression methods, such as quantization and pruning, are often deployed, especially on edge devices. In this work, we focus on layer-wise post-training quantization and pruning. Drawing connections between activation-aware weight pruning and sparse approximation problems, and motivated by the success of Iterative Hard Thresholding (IHT), we propose a unified method for Activation-aware Weight pruning and quantization via Projected gradient descent (AWP). Our experiments demonstrate that AWP outperforms state-of-the-art LLM pruning and quantization methods. Theoretical convergence guarantees of the proposed method for pruning are also provided.
Large language models (LLMs) achieve impressive performance across domains but face significant challenges when deployed on consumer-grade GPUs or personal devices such as laptops, due to high memory consumption and inference costs. Post-training quantization (PTQ) of LLMs offers a promising solution that reduces their memory footprint and decoding latency. In practice, PTQ with uniform quantization representation is favored for its efficiency and ease of deployment since uniform quantization is widely supported by mainstream hardware and software libraries. Recent studies on $\geq 2$-bit uniform quantization have led to noticeable improvements in post-quantization model performance; however, they primarily focus on quantization methodologies, while the initialization of quantization parameters is underexplored and still relies on the suboptimal Min-Max strategies. In this work, we propose NeUQI, a method devoted to efficiently determining near-optimal initial parameters for uniform quantization. NeUQI is orthogonal to prior quantization methodologies and can seamlessly integrate with them. The experiments with the LLaMA and Qwen families on various tasks demonstrate that our NeUQI consistently outperforms existing methods. Furthermore, when combined with a lightweight distillation strategy, NeUQI can achieve superior performance to PV-tuning, a much more resource-intensive approach.
Vision-Language-Action (VLA) models have shown impressive capabilities across a wide range of robotics manipulation tasks. However, their growing model size poses significant challenges for deployment on resource-constrained robotic systems. While 1-bit pretraining has proven effective for enhancing the inference efficiency of large language models with minimal performance loss, its application to VLA models remains underexplored. In this work, we present BitVLA, the first 1-bit VLA model for robotics manipulation, in which every parameter is ternary, i.e., {-1, 0, 1}. To further reduce the memory footprint of the vision encoder, we propose the distillation-aware training strategy that compresses the full-precision encoder to 1.58-bit weights. During this process, a full-precision encoder serves as a teacher model to better align latent representations. Despite the lack of large-scale robotics pretraining, BitVLA achieves performance comparable to the state-of-the-art model OpenVLA-OFT with 4-bit post-training quantization on the LIBERO benchmark, while consuming only 29.8% of the memory. These results highlight BitVLA's promise for deployment on memory-constrained edge devices. We release the code and model weights in https://github.com/ustcwhy/BitVLA.