![[2602.12338] Wireless TokenCom: RL-Based Tokenizer Agreement for Multi-User Wireless Token Communications](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.12338] Wireless TokenCom: RL-Based Tokenizer Agreement for Multi-User Wireless Token Communications
Summary
The paper presents Wireless TokenCom, a novel framework utilizing reinforcement learning for tokenizer agreement in multi-user wireless communications, enhancing efficiency and video transmission quality.
Why It Matters
As wireless networks evolve, efficient communication methods are crucial. This research addresses the challenge of tokenizer agreement in multi-user scenarios, potentially improving resource allocation and communication quality in future networks, which is vital for applications like video streaming.
Key Takeaways
- Introduces a hybrid RL framework for tokenizer agreement in wireless communications.
- Demonstrates improved semantic quality and resource efficiency over traditional methods.
- Reduces freezing events in video transmission by 68% compared to H.265.
- Addresses the complexities of multi-user communication scenarios.
- Highlights the importance of shared semantic spaces in digital communications.
Computer Science > Machine Learning arXiv:2602.12338 (cs) [Submitted on 12 Feb 2026] Title:Wireless TokenCom: RL-Based Tokenizer Agreement for Multi-User Wireless Token Communications Authors:Farshad Zeinali, Mahdi Boloursaz Mashhadi, Dusit Niyato, Rahim Tafazolli View a PDF of the paper titled Wireless TokenCom: RL-Based Tokenizer Agreement for Multi-User Wireless Token Communications, by Farshad Zeinali and 3 other authors View PDF HTML (experimental) Abstract:Token Communications (TokenCom) has recently emerged as an effective new paradigm, where tokens are the unified units of multimodal communications and computations, enabling efficient digital semantic- and goal-oriented communications in future wireless networks. To establish a shared semantic latent space, the transmitters/receivers in TokenCom need to agree on an identical tokenizer model and codebook. To this end, an initial Tokenizer Agreement (TA) process is carried out in each communication episode, where the transmitter/receiver cooperate to choose from a set of pre-trained tokenizer models/ codebooks available to them both for efficient TokenCom. In this correspondence, we investigate TA in a multi-user downlink wireless TokenCom scenario, where the base station equipped with multiple antennas transmits video token streams to multiple users. We formulate the corresponding mixed-integer non-convex problem, and propose a hybrid reinforcement learning (RL) framework that integrates a deep Q-network (DQN) for j...
