Semantic Sensing: A Task-Oriented Paradigm
arXiv:2603.29102v1 Announce Type: new Abstract: Sensing and communication are fundamental enablers of next-generation networks. While communication technologies have advanced significantly, sensing remains limited to conventional parameter estimation and is far from fully explored. Motivated by these limitations, we propose semantic sensing (SemS), a novel framework that shifts the design objective from reconstruction fidelity to semantic effective recognition. Specifically, we mathematically formulate the interaction between transmit waveforms and semantic entities, thereby establishing SemS as a semantics-oriented transceiver design. Within this architecture, we leverage the information bottleneck (IB) principle as a theoretical criterion to derive a unified objective, guiding the sensin
View PDF HTML (experimental)
Abstract:Sensing and communication are fundamental enablers of next-generation networks. While communication technologies have advanced significantly, sensing remains limited to conventional parameter estimation and is far from fully explored. Motivated by these limitations, we propose semantic sensing (SemS), a novel framework that shifts the design objective from reconstruction fidelity to semantic effective recognition. Specifically, we mathematically formulate the interaction between transmit waveforms and semantic entities, thereby establishing SemS as a semantics-oriented transceiver design. Within this architecture, we leverage the information bottleneck (IB) principle as a theoretical criterion to derive a unified objective, guiding the sensing pipeline to maximize task-relevant information extraction. To practically solve this optimization problem, we develop a deep learning (DL)-based framework that jointly designs transmit waveform parameters and receiver representations. The framework is implemented in an orthogonal frequency division multiplexing (OFDM) system, featuring a shared semantic encoder that employs a Gumbel-Softmax-based pilot selector to discretely mask task-irrelevant resources. At the receiver, we design distinct decoding architectures tailored to specific sensing objectives, comprising a 2D residual network (ResNet)-based classifier for target recognition and a correlation-driven 1D regression network for high-precision delay estimation. Numerical results demonstrate that the proposed semantic pilot design achieves superior classification accuracy and ranging precision compared to reconstruction-based baselines, particularly under constrained resource budgets.
Subjects:
Signal Processing (eess.SP)
Cite as: arXiv:2603.29102 [eess.SP]
(or arXiv:2603.29102v1 [eess.SP] for this version)
https://doi.org/10.48550/arXiv.2603.29102
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Xiaoqi Zhang [view email] [v1] Tue, 31 Mar 2026 00:42:58 UTC (1,602 KB)
Sign in to highlight and annotate this article
Conversation starters
Daily AI Digest
Get the top 5 AI stories delivered to your inbox every morning.
More about
announcearxiv
FDB Just Launched the First MCP Server for Medication Decisions
<p><em>April 1st, 2026 | MCP & Agentic AI</em></p> <p>Every week, a new AI tool promises to "transform healthcare." Most of them are polished wrappers around a general-purpose LLM with a medical disclaimer buried in the footer.</p> <p>FDB's MedProof MCP™ is not that.</p> <p>On March 31, 2026, First Databank (FDB) announced the general availability of FDB MedProof MCP™ — the first Model Context Protocol server built specifically for AI agent-driven medication decisions. Not a chatbot. Not a demo. Production infrastructure, live today, already running inside platforms that touch over 100 million patients.</p> <h2> The Problem Nobody Was Solving Correctly </h2> <p>Here's a number worth sitting with: pharmacists spend <strong>30–40% of every shift</strong> manually checking medication orde
Snapchat’s ‘Reals’ joke mocks Instagram’s many ripoffs
It's April Fools' Day, and Snap is using the occasion to "announce" that it's switching the name of Snapchat's Spotlight feed, the app's Instagram Reels-like feed of shortform vertical videos, to "Reals." In a video featuring co-founder and CEO Evan Spiegel, Snap also includes a description of "Reals" as a place where "real people share […]
Google Drive can now detect ransomware and roll back your files
Google officially announced its new anti-ransomware protections in September 2025, and the company is now making these tools available to Workspace customers using Google Drive for Desktop. The security features leverage a specially trained AI model, which has been further developed and refined over the past few months. Read Entire Article
Knowledge Map
Connected Articles — Knowledge Graph
This article is connected to other articles through shared AI topics and tags.
More in Releases
Monorepo Architecture with pnpm Workspace, Turborepo & Changesets 📦
<p>When you're developing a project with multiple packages, managing each one in its own repo can quickly turn into a nightmare. In this article, we'll set up a monorepo architecture from scratch using <strong>pnpm workspace</strong>, speed up build processes with <strong>Turborepo</strong>, and build an automated NPM publish pipeline with <strong>Changesets</strong>.</p> <h2> 🏗️ What Is a Monorepo? </h2> <p>Let's say you're building a design system. You have a core package, a theme package, and a utils package. Now imagine keeping all of these in <strong>separate repositories</strong>.</p> <p>When you fix a bug in the core package, what happens? You switch to the theme repo and update the dependency. Then you switch to the utils repo. You open separate PRs for each, wait for separate CI/
Bain’s Gross Says CEOs Get AI Wrong, Focus on Tech Over Strategy
Artificial intelligence is being misapplied by executives who treat it as a technology rollout rather than a fundamental rethink of how businesses operate, Bain Capital Managing Partner David Gross said in a Bloomberg TV interview.
PythPulse :Real-Time Crypto Anomaly Detector on Pyth Network
<p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Farnyrfipxxdx298lyri4.png" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Farnyrfipxxdx298lyri4.png" alt=" " width="800" height="450"></a>I built PythPulse for the Pyth Hackathon 2026. It monitors 38 live price feeds using Pyth's Hermes API and detects market anomalies in real-time. Live demo: <a href="https://pythpulse-2-wgig.vercel.app/" rel="noopener noreferrer">https://pythpulse-2-wgig.vercel.app/</a></p>
Discussion
Sign in to join the discussion
No comments yet — be the first to share your thoughts!