Related papers: Low-Latency and Low-Complexity MLSE for Short-Reac…

Recently, lattice reduction (LR) technique has caught great attention for multi-input multi-output (MIMO) receiver because of its low complexity and high performance. However, when the number of antennas is large, LR-aided linear detectors…

The BCJR algorithm is renowned for its optimal equalization, minimizing bit error rate (BER) over intersymbol interference (ISI) channels. However, its complexity grows exponentially with the channel memory, posing a significant…

The focus of this paper is a proof of concept, machine learning (ML) pipeline that extracts heart rate from pressure sensor data acquired on low-power edge devices. The ML pipeline consists an upsampler neural network, a signal quality…

LLM-based ASR overcomes multilingual data scarcity by projecting speech representations into the LLM space to leverage its robust semantic and reasoning capabilities. However, while previous approaches typically enhance performance by…

Speculative decoding accelerates Large Language Model (LLM) inference by employing a small speculative model (SSM) to generate multiple candidate tokens and verify them using the LLM in parallel. This technique has been widely integrated…

While variance reduction methods have shown great success in solving large scale optimization problems, many of them suffer from accumulated errors and, therefore, should periodically require the full gradient computation. In this paper, we…

In orthogonal frequency division multiplexing (OFDM), accurate channel estimation is crucial. Classical signal processing-based approaches, such as linear minimum mean-squared error (LMMSE) estimation, often require second-order statistics…

Multiple-input multiple-output (MIMO) systems have been widely acclaimed in order to provide high data rates. Recently Lattice Reduction (LR) aided detectors have been proposed to achieve near Maximum Likelihood (ML) performance with low…

In this study, we revisit the performance analysis of distributed beamforming architectures in dense user-centric cell-free massive multiple-input multiple-output (mMIMO) systems in line-of-sight (LoS) scenarios. By incorporating a recently…

This paper studies the error rate performance and low-complexity receiver design for zero-padded affine frequency division multiplexing (ZP-AFDM) systems. By exploiting the unique ZP-aided lower triangular structure of the time domain (TD)…

In space applications, hardware (HW) implementation is made more expensive not only by the levels of performance required, but also by complex and rigorous HW qualification tests. Reducing qualification cost and time is thus a key design…

Since Tse and Verdu proved that the global maximum likelihood (GML) detector achieves unit asymptotic multiuser efficiency (AME) in the limit of large random spreading (LRS) CDMA, no suboptimal detector has been found to achieve unit AME.…

MoE-PEFT methods combine Mixture of Experts with parameter-efficient fine-tuning for multi-task adaptation, but require separate adapters per expert causing trainable parameters to scale linearly with expert count and limiting applicability…

We investigate the integration of Large Language Models (LLMs) into query encoders to improve dense retrieval without increasing latency and cost, by circumventing the dependency on LLMs at inference time. SoftQE incorporates knowledge from…

A lattice reduction is an algorithm that transforms the given basis of the lattice to another lattice basis such that problems like finding a shortest vector and closest vector become easier to solve. Some of the famous lattice reduction…

Federated inference enhances LLM performance in edge computing through weighted averaging of distributed model predictions. However, autoregressive LLM inference requires frequent full-model forward passes across workers, severely limiting…

Linear minimum mean square error (LMMSE) receivers are often applied in practical communication scenarios for single-input-multiple-output (SIMO) systems owing to their low computational complexity and competitive performance. However,…

In the transmission of digital data at a relatively high rate over a particular band limited channel, it is normally necessary to employ an equalizer at the receiver in order to correct the signal distortion introduced by the channel .ISI…

Mixture-of-Experts (MoE) models are typically pre-trained with explicit load-balancing constraints to ensure statistically balanced expert routing. Despite this, we observe that even well-trained MoE models exhibit significantly imbalanced…

In this paper, first, a hardware-friendly pruning algorithm for reducing energy consumption and improving the speed of Long Short-Term Memory (LSTM) neural network accelerators is presented. Next, an FPGA-based platform for efficient…