Pub Date : 2024-11-04DOI: 10.1109/LNET.2024.3490954
Changshi Zhou;Weiqi Liu;Tao Han;Nirwan Ansari
From AI-assisted art creation to large language model (LLM)-powered ChatGPT, AI-generated contents and services are becoming a transforming force. It calls for the telecom industry to embrace the prospects of AIGC services and face the unique challenges posed by incorporating generative model services into the AI-native 6G wireless network paradigm. We propose enabling AIGC inference services on mobile devices by optimizing MEC-device computing offloading, through which AIGC task latency is minimized by reinforcement learning based policy agent in a computing resource constrained and bandwidth limited wireless environment. Simulation results are presented to demonstrate the performance advantage.
{"title":"Deploying On-Device AIGC Inference Services in 6G via Optimal MEC-Device Offloading","authors":"Changshi Zhou;Weiqi Liu;Tao Han;Nirwan Ansari","doi":"10.1109/LNET.2024.3490954","DOIUrl":"https://doi.org/10.1109/LNET.2024.3490954","url":null,"abstract":"From AI-assisted art creation to large language model (LLM)-powered ChatGPT, AI-generated contents and services are becoming a transforming force. It calls for the telecom industry to embrace the prospects of AIGC services and face the unique challenges posed by incorporating generative model services into the AI-native 6G wireless network paradigm. We propose enabling AIGC inference services on mobile devices by optimizing MEC-device computing offloading, through which AIGC task latency is minimized by reinforcement learning based policy agent in a computing resource constrained and bandwidth limited wireless environment. Simulation results are presented to demonstrate the performance advantage.","PeriodicalId":100628,"journal":{"name":"IEEE Networking Letters","volume":"6 4","pages":"232-236"},"PeriodicalIF":0.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143388625","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-25DOI: 10.1109/LNET.2024.3486260
Mohammad Ghassemi;Han Zhang;Ali Afana;Akram Bin Sediq;Melike Erol-Kantarci
Beam management is an important technique to improve signal strength and reduce interference in wireless communication systems. Recently, there has been increasing interest in using diverse sensing modalities for beam management. However, it remains a big challenge to process multi-modal data efficiently and extract useful information. On the other hand, the recently emerging multi-modal transformer (MMT) is a promising technique that can process multi-modal data by capturing long-range dependencies. While MMT is highly effective in handling multi-modal data and providing robust beam management, integrating reinforcement learning (RL) further enhances their adaptability in dynamic environments. In this letter, we propose a two-step beam management method by combining MMT with RL for dynamic beam index prediction. In the first step, we divide available beam indices into several groups and leverage MMT to process diverse data modalities to predict the optimal beam group. In the second step, we employ RL for fast beam decision-making within each group, which in return maximizes throughput. Our proposed framework is tested on a 6G dataset. In this testing scenario, it achieves higher beam prediction accuracy and system throughput compared to both the MMT-only based method and the RL-only based method.
{"title":"Multi-Modal Transformer and Reinforcement Learning-Based Beam Management","authors":"Mohammad Ghassemi;Han Zhang;Ali Afana;Akram Bin Sediq;Melike Erol-Kantarci","doi":"10.1109/LNET.2024.3486260","DOIUrl":"https://doi.org/10.1109/LNET.2024.3486260","url":null,"abstract":"Beam management is an important technique to improve signal strength and reduce interference in wireless communication systems. Recently, there has been increasing interest in using diverse sensing modalities for beam management. However, it remains a big challenge to process multi-modal data efficiently and extract useful information. On the other hand, the recently emerging multi-modal transformer (MMT) is a promising technique that can process multi-modal data by capturing long-range dependencies. While MMT is highly effective in handling multi-modal data and providing robust beam management, integrating reinforcement learning (RL) further enhances their adaptability in dynamic environments. In this letter, we propose a two-step beam management method by combining MMT with RL for dynamic beam index prediction. In the first step, we divide available beam indices into several groups and leverage MMT to process diverse data modalities to predict the optimal beam group. In the second step, we employ RL for fast beam decision-making within each group, which in return maximizes throughput. Our proposed framework is tested on a 6G dataset. In this testing scenario, it achieves higher beam prediction accuracy and system throughput compared to both the MMT-only based method and the RL-only based method.","PeriodicalId":100628,"journal":{"name":"IEEE Networking Letters","volume":"6 4","pages":"222-226"},"PeriodicalIF":0.0,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143388620","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-24DOI: 10.1109/LNET.2024.3486194
Yinyu Wu;Xuhui Zhang;Jinke Ren;Huijun Xing;Yanyan Shen;Shuguang Cui
Recently, the integration of mobile edge computing (MEC) and generative artificial intelligence (GAI) technology has given rise to a new area called mobile edge generation and computing (MEGC), which offers mobile users heterogeneous services such as task computing and content generation. In this letter, we investigate the joint communication, computation, and the AIGC resource allocation problem in an MEGC system. A latency minimization problem is first formulated to enhance the quality of service for mobile users. Due to the strong coupling of the optimization variables, we propose a new deep reinforcement learning-based algorithm to solve it efficiently, Numerical results demonstrate that the proposed algorithm can achieve lower latency than several baseline algorithms.
{"title":"Latency-Aware Resource Allocation for Mobile Edge Generation and Computing via Deep Reinforcement Learning","authors":"Yinyu Wu;Xuhui Zhang;Jinke Ren;Huijun Xing;Yanyan Shen;Shuguang Cui","doi":"10.1109/LNET.2024.3486194","DOIUrl":"https://doi.org/10.1109/LNET.2024.3486194","url":null,"abstract":"Recently, the integration of mobile edge computing (MEC) and generative artificial intelligence (GAI) technology has given rise to a new area called mobile edge generation and computing (MEGC), which offers mobile users heterogeneous services such as task computing and content generation. In this letter, we investigate the joint communication, computation, and the AIGC resource allocation problem in an MEGC system. A latency minimization problem is first formulated to enhance the quality of service for mobile users. Due to the strong coupling of the optimization variables, we propose a new deep reinforcement learning-based algorithm to solve it efficiently, Numerical results demonstrate that the proposed algorithm can achieve lower latency than several baseline algorithms.","PeriodicalId":100628,"journal":{"name":"IEEE Networking Letters","volume":"6 4","pages":"237-241"},"PeriodicalIF":0.0,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143388485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-17DOI: 10.1109/LNET.2024.3482295
Yantao Li;Shijun Ling;Hongyu Huang;Feno H. Rabevohitra
Contrastive unsupervised learning has made significant progress, but there is still potential for improvement by capturing finer details in input data. In this letter, we present PGMoCo, a Progressive Growth-based Momentum Contrast framework for unsupervised representative learning in classification tasks. PGMoCo begins by learning the overall distribution of samples at a coarse scale and progressively refines the representation by incorporating increasingly finer details. PGMoCo consists of data augmentation, progressive growth, an alternative multilayer perceptron (MLP) head, and a loss function. First, PGMoCo applies transformation-based data augmentation to the input samples. Then, it progressively learns features at multiple scales, uses an alternative MLP head to project latent representations into a contrastive loss space, and finally employs a specialized loss function to classify the samples. We evaluate PGMoCo on three datasets: CIFAR-10 and PolyU Palmprint (image classification) and H-MOG (person identification). PGMoCo achieves classification accuracies of 86.76% on CIFAR-10, 95.94% on PolyU Palmprint, and 80.10% on H-MOG, outperforming existing state-of-the-art methods.
{"title":"Progressive Growth-Based Momentum Contrast for Unsupervised Representative Learning in Classification Tasks","authors":"Yantao Li;Shijun Ling;Hongyu Huang;Feno H. Rabevohitra","doi":"10.1109/LNET.2024.3482295","DOIUrl":"https://doi.org/10.1109/LNET.2024.3482295","url":null,"abstract":"Contrastive unsupervised learning has made significant progress, but there is still potential for improvement by capturing finer details in input data. In this letter, we present PGMoCo, a Progressive Growth-based Momentum Contrast framework for unsupervised representative learning in classification tasks. PGMoCo begins by learning the overall distribution of samples at a coarse scale and progressively refines the representation by incorporating increasingly finer details. PGMoCo consists of data augmentation, progressive growth, an alternative multilayer perceptron (MLP) head, and a loss function. First, PGMoCo applies transformation-based data augmentation to the input samples. Then, it progressively learns features at multiple scales, uses an alternative MLP head to project latent representations into a contrastive loss space, and finally employs a specialized loss function to classify the samples. We evaluate PGMoCo on three datasets: CIFAR-10 and PolyU Palmprint (image classification) and H-MOG (person identification). PGMoCo achieves classification accuracies of 86.76% on CIFAR-10, 95.94% on PolyU Palmprint, and 80.10% on H-MOG, outperforming existing state-of-the-art methods.","PeriodicalId":100628,"journal":{"name":"IEEE Networking Letters","volume":"7 1","pages":"31-35"},"PeriodicalIF":0.0,"publicationDate":"2024-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143645189","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-14DOI: 10.1109/LNET.2024.3480149
Ahmet Gunhan Aydin;Haris Vikalo
The pursuit of enhanced interactive visual experiences has created growing interest in 360-degree video streaming. However, transmitting such content requires significant bandwidth compared to conventional planar video, motivating a search for effective bandwidth optimization strategies. A promising approach involves predicting viewport and prioritizing transmission of the regions of interest at higher quality. The existing methods for viewport prediction rely on sophisticated neural networks hosted on servers and face major bandwidth and latency challenges. This letter proposes a hierarchical approach to viewport prediction that leverages a small model on edge devices and offloads to the server only the most challenging tasks. The offloading algorithm relies on rate control to maximize the performance while meeting resource constraints, presenting a novel solution to bandwidth-efficient viewport prediction for 360-degree video streaming.
{"title":"Viewport Prediction via Adaptive Edge Offloading","authors":"Ahmet Gunhan Aydin;Haris Vikalo","doi":"10.1109/LNET.2024.3480149","DOIUrl":"https://doi.org/10.1109/LNET.2024.3480149","url":null,"abstract":"The pursuit of enhanced interactive visual experiences has created growing interest in 360-degree video streaming. However, transmitting such content requires significant bandwidth compared to conventional planar video, motivating a search for effective bandwidth optimization strategies. A promising approach involves predicting viewport and prioritizing transmission of the regions of interest at higher quality. The existing methods for viewport prediction rely on sophisticated neural networks hosted on servers and face major bandwidth and latency challenges. This letter proposes a hierarchical approach to viewport prediction that leverages a small model on edge devices and offloads to the server only the most challenging tasks. The offloading algorithm relies on rate control to maximize the performance while meeting resource constraints, presenting a novel solution to bandwidth-efficient viewport prediction for 360-degree video streaming.","PeriodicalId":100628,"journal":{"name":"IEEE Networking Letters","volume":"7 1","pages":"21-25"},"PeriodicalIF":0.0,"publicationDate":"2024-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143645190","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-14DOI: 10.1109/LNET.2024.3479914
Monolina Dutta;Anoop Thomas;Frank Y. Li
Traditional single server based coded caching may face server saturation and service vulnerability problems. In this letter, we integrate decentralized coded caching with a multi-server architecture comprising both data and parity servers. For file distribution in this network, a method referred to as file stripping is adopted, and a novel file delivery scheme is proposed. Closed-form expressions for the total transmission rate achieved by this scheme are derived, considering all the operational servers along with the worst-case transmission rate amongst these servers. Additionally, a comparative analysis between the proposed scheme and the conventional decentralized coded caching scheme is presented. The simulation results demonstrate the viability of our proposed scheme.
{"title":"Decentralized Coded Caching With Distributed Storage Across Data and Parity Servers","authors":"Monolina Dutta;Anoop Thomas;Frank Y. Li","doi":"10.1109/LNET.2024.3479914","DOIUrl":"https://doi.org/10.1109/LNET.2024.3479914","url":null,"abstract":"Traditional single server based coded caching may face server saturation and service vulnerability problems. In this letter, we integrate decentralized coded caching with a multi-server architecture comprising both data and parity servers. For file distribution in this network, a method referred to as file stripping is adopted, and a novel file delivery scheme is proposed. Closed-form expressions for the total transmission rate achieved by this scheme are derived, considering all the operational servers along with the worst-case transmission rate amongst these servers. Additionally, a comparative analysis between the proposed scheme and the conventional decentralized coded caching scheme is presented. The simulation results demonstrate the viability of our proposed scheme.","PeriodicalId":100628,"journal":{"name":"IEEE Networking Letters","volume":"7 1","pages":"26-30"},"PeriodicalIF":0.0,"publicationDate":"2024-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143645188","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-07DOI: 10.1109/LNET.2024.3475269
Faris B. Mismar;Aliye Özge Kaya
This letter proposes the use of deep autoencoders to compress the channel information in a massive multiple input and multiple output (MIMO) system. Although autoencoders perform lossy compression, they still have adequate usefulness when applied to massive MIMO system channel state information (CSI) compression. To demonstrate their impact on the CSI, we measure the performance of the system under two different channel models for different compression ratios. We disclose a few practical considerations in using autoencoders for this propose. We show through simulation that the run-time complexity of this deep autoencoder is irrelative to the compression ratio and thus an adaptive compression rate is feasible with an optimal compression ratio depending on the channel model and the signal to noise ratio.
{"title":"Adaptive Compression of Massive MIMO Channel State Information With Deep Learning","authors":"Faris B. Mismar;Aliye Özge Kaya","doi":"10.1109/LNET.2024.3475269","DOIUrl":"https://doi.org/10.1109/LNET.2024.3475269","url":null,"abstract":"This letter proposes the use of deep autoencoders to compress the channel information in a massive multiple input and multiple output (MIMO) system. Although autoencoders perform lossy compression, they still have adequate usefulness when applied to massive MIMO system channel state information (CSI) compression. To demonstrate their impact on the CSI, we measure the performance of the system under two different channel models for different compression ratios. We disclose a few practical considerations in using autoencoders for this propose. We show through simulation that the run-time complexity of this deep autoencoder is irrelative to the compression ratio and thus an adaptive compression rate is feasible with an optimal compression ratio depending on the channel model and the signal to noise ratio.","PeriodicalId":100628,"journal":{"name":"IEEE Networking Letters","volume":"6 4","pages":"267-271"},"PeriodicalIF":0.0,"publicationDate":"2024-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143388646","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：