In multi-cell wireless networks, providing intelligent services via federated learning (FL) becomes more challenging due to multi-level distribution shifts across clients and regions, as well as additional communication delays among edge and cloud servers. To address these issues, we propose SplitOMC, a split learning framework that integrates overlapping-area clients and a multi-exit neural architecture to jointly handle (i) client-preferred, (ii) out-of-preference, and (iii) out-of-region tasks. By strategically leveraging clients in overlapping regions, SplitOMC accelerates training without excessive backhaul communication, while maintaining both personalization and generalization. We theoretically analyze the convergence behavior of the proposed algorithm, ensuring performance stability under heterogeneous data and communication conditions. Extensive experiments on MNIST, CIFAR-10/100, and a real-world Jetson Nano testbed demonstrate that SplitOMC consistently achieves faster training and inference with improved accuracy compared to state-of-the-art methods. In particular, the framework shows robustness in resource-constrained and unstable network environments, highlighting its practical value for next-generation wireless intelligent services.

Instruction-following large language models (LLMs), such as ChatGPT, have become widely popular among everyday users.However, these models inadvertently disclose private, sensitive information to their users, underscoring the need for machine unlearning techniques to remove selective information from the models.While prior work has focused on forgetting small, random subsets of training data at the instance-level, we argue that real-world scenarios often require the removal of an entire user data, which may require a more careful maneuver.In this study, we explore entitylevel unlearning, which aims to erase all knowledge related to a target entity while preserving the remaining model capabilities.To address this, we introduce OPT-OUT, an optimal transport-based unlearning method that utilizes the Wasserstein distance from the model's initial parameters to achieve more effective and fine-grained unlearning.We also present the first Entity-Level Unlearning Dataset (ELUDe) designed to evaluate entity-level unlearning.Our empirical results demonstrate that OPT-OUT surpasses existing methods, establishing a new standard for secure and adaptable LLMs that can accommodate user data removal requests without the need for full retraining. 1

The rapid surge in ultra-high-definition video streaming demand in next-generation mobile networks calls for intelligent content delivery technologies to ensure high service quality under dynamic and unpredictable wireless channel conditions. In this paper, we propose a novel two-timescale joint optimization framework for user scheduling and video delivery that does not rely on channel state information (CSI) and utilizes only user mobility and queueing system information. Specifically, at the small timescale, queue-driven adaptive bitrate control is performed based on user mobility to effectively reduce playback stalls and quality fluctuations. At the large timescale, user-to-node scheduling is optimized to maintain queue stability. To solve the joint problem, we integrate Lyapunov optimization with a Deep Q-Network (DQN)-based reinforcement learning approach. Simulation results demonstrate that the proposed method outperforms existing schemes in terms of playback stall rate, average bitrate, and quality fluctuation.

The demand for intelligent services at the network edge has introduced several research challenges. One is the need for a machine learning architecture that achieves personalization (to individual clients) and generalization (to unseen data) properties concurrently across different applications. Another is the need for an inference strategy that can satisfy network resource and latency constraints during testing-time. Existing techniques in federated learning have encountered a steep trade-off between personalization and generalization, and have not explicitly considered the resource requirements during the inference-stage. In this paper, we propose SplitGP, a joint edge-AI training and inference strategy that simultaneously captures generalization/personalization for efficient inference across resource-constrained clients. The training process of SplitGP is based on federated split learning, with the key idea of optimizing the client-side model to have personalization capability tailored to its main task, while training the server-side model to have generalization capability for handling out-of-distribution tasks. During testing-time, each client selectively offloads inference tasks to the server based on the uncertainty threshold tunable based on network resource availability. Through formal convergence analysis and inference time analysis, we provide guidelines on the selection of key meta-parameters in SplitGP. Experimental results confirm the advantage of SplitGP over existing baselines.

In order to build realistic metaverse systems, enabling high synchronization between physical-space and virtual meta-space is essentially required. For this purpose, this paper proposes a novel system-wide coordination algorithm for high synchronization under characteristics (i.e., highly realistic meta-space construction under the constraints of physical-space). The proposed algorithm consists of the following three stages. The first stage is quantum multi-agent reinforcement learning (QMARL)-based scheduling for low-delay temporal-synchronization using differentiated age-of-information (AoI) during data gathering in physical-space by observers for meta-space construction. This is beneficial for scalability according to action dimension reduction in reinforcement learning computation. The second stage is for creating virtual contents under delay constraints in meta-space based on the gathered data. When rendering regions that have received more user attention, avatar-popularity is considered for spatio-synchronization. Thus, a stabilized control mechanism is designed for time-average reality quality maximization for each region. The last stage is for caching based on avatar-popularity and AoI which can be helpful in constructing low-delay realistic meta-space. Furthermore, the concept of AoI is divided into two separate sub-concepts of physical AoI and virtual AoI such that the AoI in virtual meta-space can be thoroughly implemented.

Despite the extensive applications of relation extraction (RE) tasks in various domains, little has been explored in the historical context, which contains promising data across hundreds and thousands of years. To promote the historical RE research, we present HistRED constructed from Yeonhaengnok. Yeonhaengnok is a collection of records originally written in Hanja, the classical Chinese writing, which has later been translated into Korean. HistRED provides bilingual annotations such that RE can be performed on Korean and Hanja texts. In addition, HistRED supports various self-contained subtexts with different lengths, from a sentence level to a document level, supporting diverse context settings for researchers to evaluate the robustness of their RE models. To demonstrate the usefulness of our dataset, we propose a bilingual RE model that leverages both Korean and Hanja contexts to predict relations between entities. Our model outperforms monolingual baselines on HistRED, showing that employing multiple language contexts supplements the RE predictions.

Motivated by the features of video content that can have multiple versions with different qualities and be divided into several chunks, this paper proposes a caching and transcoding scheme for individual video chunks in the hierarchical wireless network. The proposed policy determines which contents, what quality, and how many chunks to be cached and transcoded depending on different cache sizes, computing resources, and communication overheads for different tiers. Considering adaptive quality selection at the transmitter end, we formulate the expected quality maximization problem, while guaranteeing delay-constrained video delivery. The proposed scheme controls tradeoffs between the following metrics: 1) transcoding and transmission time, 2) content quality and diversity of caching, and 3) caching more fractions and caching more contents. The simulation results show that the proposed scheme provides an improved content quality while guaranteeing the delay constraint, compared to traditional caching which stores the entire files only.

A fundamental challenge to providing edge-AI services is the need for a machine learning (ML) model that achieves personalization (i.e., to individual clients) and generalization (i.e., to unseen data) properties concurrently. Existing techniques in federated learning (FL) have encountered a steep tradeoff between these objectives and impose large computational requirements on edge devices during training and inference. In this paper, we propose SplitGP, a new split learning solution that can simultaneously capture generalization and personalization capabilities for efficient inference across resource-constrained clients (e.g., mobile/IoT devices). Our key idea is to split the full ML model into client-side and server-side components, and impose different roles to them: the client-side model is trained to have strong personalization capability optimized to each client’s main task, while the server-side model is trained to have strong generalization capability for handling all clients’ out-of-distribution tasks. We analytically characterize the convergence behavior of SplitGP, revealing that all client models approach stationary points asymptotically. Further, we analyze the inference time in SplitGP and provide bounds for determining model split ratios. Experimental results show that SplitGP outperforms existing baselines by wide margins in inference time and test accuracy for varying amounts of out-of-distribution samples.

A simultaneous wireless information and power transfer (SWIPT)-enabled nonorthogonal multiple access (NOMA) system has been recognized as a promising technology for enabling the Industrial Internet of Things (IIoT), as it extends the lifetime of small battery-driven sensors. However, its energy efficiency significantly decreases with distance from the central controller. This article proposes a framework for applying SWIPT-aided NOMA system to a distributed antenna system (DAS) to improve the spectral efficiency and energy efficiency of the IIoT. Moreover, we jointly optimize the power allocation for the NOMA signaling and power splitting for SWIPT in the DAS to maximize energy efficiency while reaching the minimum harvested energy and data rate requirements. Owing to the nonconvexity of the joint optimization problem, we divide the optimization problem into subproblems and present an iterative algorithm for finding the optimal solution. To verify the effectiveness of the proposed method, extensive simulation results are presented, and we show improvements in energy efficiency relative to conventional schemes. Particularly, the proposed SWIPT-NOMA-DAS system provides energy efficiency of more than five times that of SWIPT-NOMA without DAS, and more than 10% improvement compared with SWIPT-OMA-DAS.

Hojun Cho, Dohee Kim, Seungwoo Ryu, ChaeHun Park, Hyungjong Noh, Jeong-in Hwang, Minseok Choi, Edward Choi, Jaegul Choo. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing. 2022.

We consider a scenario that utilizes road side units (RSUs) as distributed caches in connected vehicular networks. The goal of the use of caches in our scenario is for rapidly providing contents to connected vehicles under various traffic conditions. During this operation, due to the rapidly changed road environment and user mobility, the concept of age-of-information (AoI) is considered for (1) updating the cached information as well as (2) maintaining the freshness of cached information. The frequent updates of cached information maintain the freshness of the information at the expense of network resources. Here, the frequent updates increase the number of data transmissions between RSUs and MBS; and thus, it increases system costs, consequently. Therefore, the tradeoff exists between the AoI of cached information and the system costs. Based on this observation, the proposed algorithm in this paper aims at the system cost reduction which is fundamentally required for content delivery while minimizing the content AoI, based on Markov Decision Process (MDP) and Lyapunov optimization.

This paper proposes a cooperative and dynamic quality adaptation scheme for delay-sensitive video streaming between the transmitter and the receiver. We present a novel adaptive super-resolution (SR) technique that adaptively controls the quality enhancement rate and computation time. Due to the capability of enhancing the quality of video chunks at the user device side, the transmitter can aggressively transcode video chunks to reduce the delivery latency and power consumption. Also, adaptive SR can control the tradeoff between playback stall rate and CPU consumption of the user device. Simulation results verify the performance of adaptive SR and show that the proposed video delivery scheme is very good to balance tradeoff among the following performance metrics for online video services: 1) playback stall rate, 2) average quality measure, 3) transmission power, and 4) CPU consumption of the user device.

In this demonstration, we present AB9 SoC system, a single-chip solution for AI application. It provides the reconfigurable and programmable architecture to support the general computations for a variety of neural networks. The AB9 SoC is implemented using TSMC 28-nm process technology with a chip size of 17×23 mm2and 1GHz operating frequency.

While federated learning (FL) allows efficient model training with local data at edge devices, among major issues still to be resolved are: slow devices known as stragglers and malicious attacks launched by adversaries. While the presence of both of these issues raises serious concerns in practical FL systems, no known schemes or combinations of schemes effectively address them at the same time. We propose Sageflow, staleness-aware grouping with entropy-based filtering and loss-weighted averaging, to handle both stragglers and adversaries simultaneously. Model grouping and weighting according to staleness (arrival delay) provides robustness against stragglers, while entropy-based filtering and loss-weighted averaging, working in a highly complementary fashion at each grouping stage, counter a wide range of adversary attacks. A theoretical bound is established to provide key insights into the convergence behavior of Sageflow. Extensive experimental results show that Sageflow outperforms various existing methods aiming to handle stragglers/adversaries.

We consider federated learning (FL) with multiple wireless edge servers having their own local coverage. We focus on speeding up training in this increasingly practical setup. Our key idea is to utilize the clients located in the overlapping coverage areas among adjacent edge servers (ESs); in the model-downloading stage, the clients in the overlapping areas receive multiple models from different ESs, take the average of the received models, and then update the averaged model with their local data. These clients send their updated model to multiple ESs by broadcasting, which acts as bridges for sharing the trained models between servers. Even when some ESs are given biased datasets within their coverage regions, their training processes can be assisted by adjacent servers through the clients in their overlapping regions. As a result, the proposed scheme does not require costly communications with the central cloud server (located at the higher tier of edge servers) for model synchronization, significantly reducing the overall training time compared to the conventional cloud-based FL systems. Extensive experimental results show remarkable performance gains of our scheme compared to existing methods. Our design targets latency-sensitive applications where edge-based FL is essential, e.g., when a number of connected cars/drones must cooperate (via FL) to quickly adapt to dynamically changing environments.

Wireless caching networks have been extensively researched as a promising technique for supporting the massive data traffic of multimedia services. Many of the existing studies on real-data traffic have shown that users of a multimedia service consecutively request multiple contents and this sequence is strongly dependent on the related list of the first content and/or the top referrer in the category. This paper thus introduces the notion of “temporary preference”, characterizing the behavior of users who are highly likely to request the next content from a certain target category (i.e., related content list). Based on this observation, this paper proposes both probabilistic caching and dynamic delivery policies for categorized contents and consecutive user demands. The proposed caching scheme maximizes the minimum of the cache hit rates for all users. In the delivery phase, a dynamic helper association policy for receiving multiple contents in a row is designed to reduce the delivery latency. By comparing with the content placement optimized for one-shot requests, numerical results verify the effects of categorized contents and consecutive user demands on the proposed caching and delivery policies.

This paper proposes a user scheduling and power allocation method for content delivery in wireless caching helper networks without any stringent constraint on the interference model. For supporting delay-sensitive and time-varying user demands, the actual delivery quantity of the requested content should be dynamically controlled by advanced scheduling and power allocation. In addition, it is difficult for a central unit to control the content delivery due to a lack of knowledge of the entire time-varying network; therefore, a belief-propagation (BP)-based algorithm that facilitates distributed decisions on user scheduling and power allocation at every caching helper is presented. The proposed delivery scheme maximizes power efficiency while limiting the average delay of user request satisfactions by managing interference among users well. Simulation results show that the proposed scheme provides almost the same delay performance as the exhaustively found optimal one at the expense of little power consumption.

In wireless caching networks, a user generally has a concrete purpose of consuming contents in a certain preferred category, and requests multiple contents in sequence. While most existing research on wireless caching and delivery has focused only on one-shot requests, the popularity distribution of contents requested consecutively is definitely different from the one-shot request and has been not considered. Also, especially from the perspective of the service provider, it is advantageous for users to consume as many contents as possible. Thus, this paper proposes two cache allocation policies for categorized contents and consecutive user demands, which maximize 1) the cache hit rate and 2) the number of consecutive content consumption, respectively. Numerical results show how categorized contents and consecutive content requests have impacts on the cache allocation.

In wireless caching networks, the design of the content delivery method must consider random user requests, caching states, network topology, and interference management. In this article, we establish a general framework for content delivery in wireless caching networks without stringent assumptions that restrict the network structure and interference model. Based on the framework, we propose a dynamic and distributed link scheduling and power allocation scheme for content delivery that is assisted by belief-propagation (BP) algorithms. The proposed scheme achieves three critical purposes of wireless caching networks: 1) limiting the delay of user request satisfactions, 2) maintaining the power efficiency of caching nodes, and 3) managing interference among users. In addition, we address the intrinsic problem of the BP algorithm in our network model, proposing a matching algorithm for one-to-one link scheduling. Simulation results show that the proposed scheme provides almost the same delay performance as the optimal scheme found through an exhaustive search at the expense of a little additional power consumption and does not require a clustering method and orthogonal resources in a large-scale D2D network.

In wireless caching networks, each user generally consumes more than one content in a row, and the number of consecutive demands could vary for different users. In addition, popular contents are usually classified into several categories. In this case for consecutive user demands, the content popularity model largely depends on the previously consumed contents, i.e., contents that belong to the same category as the previously consumed content would be highly popular. Based on this observation, this paper proposes an optimal probabilistic caching policy for consecutive user demands in categorized contents. The proposed caching scheme maximizes the minimum of the success probabilities for content delivery of all users when individual users request different numbers of contents in a row. Comparing with the content placement optimized for one-shot request, intensive numerical results verify the impacts of categorized contents and consecutive user demands on the caching policy.

In this paper, we propose a joint dynamic power control and user pairing algorithm for power-efficient and delay-constrained hybrid multiple access systems. In a hybrid multiple access system, user pairing determines whether the transmitter serves as a certain user by orthogonal multiple access (OMA) or non-orthogonal multiple access (NOMA). The proposed optimization framework minimizes the long-term time-average transmit power expenditure while reducing the queuing delay and guaranteeing the minimum time-average data rates. The proposed technique observes both channel and queue state information and adjusts queue backlogs to avoid an excessive queueing delay by appropriate user pairing and power allocation. Furthermore, the flexible use of resources is captured in the proposed algorithm by employing NOMA. The data-intensive simulation results show that the proposed scheme for power allocation and user scheduling achieves a balance among multiple performance goals, i.e., power efficiency, queueing delay, and data rate.

This paper proposes a video delivery strategy for dynamic streaming services which maximizes time-average streaming quality under a playback delay constraint in wireless caching networks. The network where popular videos encoded by scalable video coding are already stored in randomly distributed caching nodes is considered under adaptive video streaming concepts, and distance-based interference management is investigated in this paper. In this network model, a streaming user makes delay-constrained decisions depending on stochastic network states: 1) caching node for video delivery, 2) video quality, and 3) the quantity of video chunks to receive. Since wireless link activation for video delivery may introduce delays, different timescales for updating caching node association, video quality adaptation, and chunk amounts are considered. After associating with a caching node for video delivery, the streaming user chooses combinations of quality and chunk amounts in the small timescale. The dynamic decision making process for video quality and chunk amounts at each slot is modeled using Markov decision process, and the caching node decision is made based on the framework of Lyapunov optimization. Our intensive simulations verify that the proposed video delivery algorithm works reliably and also can control the tradeoff between video quality and playback latency.

This paper considers one-hop device-to-device-assisted wireless caching networks that cache video files of varying quality levels, with the assumption that the base station can control the video quality but cache-enabled devices cannot. Two problems arise in such a caching network: file placement problem and node association problem. This paper suggests a method to cache videos of different qualities, and thus of varying file sizes, by maximizing the sum of video quality measures that users can enjoy. There exists an interesting tradeoff between video quality and video diversity, i.e., the ability to provision diverse video files. By caching high-quality files, the cache-enabled devices can provide high-quality video, but cannot cache a variety of files. Conversely, when the device caches various files, it cannot provide a good quality for file-requesting users. In addition, when multiple devices cache the same file but their qualities are different, advanced node association is required for file delivery. This paper proposes a node association algorithm that maximizes time-averaged video quality for multiple users under a playback delay constraint. In this algorithm, we also consider request collision, the situation where several users request files from the same device at the same time, and we propose two ways to cope with the collision: scheduling of one user and non-orthogonal multiple access. Simulation results verify that the proposed caching method and the node association algorithm work reliably.

In this paper, we propose bi-directional cooperative non-orthogonal multiple access (NOMA). Compared to conventional NOMA, the main contributions of bi-directional cooperative NOMA can be explained in two directions: 1) the proposed NOMA system is still efficient when the channel gains of scheduled users are almost the same and 2) the proposed NOMA system operates well without accurate channel-state information at the base station. In a two-user scenario, the closed-form ergodic capacity of bi-directional cooperative NOMA is derived, and it is proven to be better than those of other techniques. Based on the ergodic capacity, the algorithms to find optimal power allocations maximizing the user fairness and sum rate are presented. Outage probability is also derived, and we show that bi-directional cooperative NOMA achieves a power gain over uni-directional cooperative NOMA and a diversity gain over non-cooperative NOMA and orthogonal multiple access (OMA). We finally extend the bi-directional cooperative NOMA to a multiuser model. The analysis of ergodic capacity and outage probability in a two-user scenario is numerically verified. Also, simulation results show that bi-directional cooperative NOMA provides better data rates than the existing NOMA schemes as well as OMA in a multiuser scenario.

We address general filtering problems on the Euclidean group SE(3). We first generalize, to stochastic nonlinear systems evolving on SE(3)9 the particle filter of Liu and West (2001) for simultaneously estimating the state and covariance. The filter is constructed in a coordinate-invariant way, and explicitly takes into account the geometry of SE(3) and P(n)9 the space of symmetric positive definite matrices. An experimental case study involving vision-based robot end-effector pose estimation is also presented.

A widely used approach to image registration involves finding the general linear transformation that maximizes the mutual information between two images, with the transformation being rigid-body [i.e., belonging to SE(3)] or volume-preserving [i.e., belonging to SL(3)]. In this paper, we present coordinate-invariant, geometric versions of the Nelder-Mead optimization algorithm on the groups SL(3), SE(3), and their various subgroups, that are applicable to a wide class of image registration problems. Because the algorithms respect the geometric structure of the underlying groups, they are numerically more stable, and exhibit better convergence properties than existing local coordinate-based algorithms. Experimental results demonstrate the improved convergence properties of our geometric algorithms.

We present a low-power architectural MPEG-4 part-10 AVC/H.264 video and MPEG-4 BSAC audio decoder chip capable of delivering high-quality and high-compression in wireless multimedia applications such as DMB (digital multimedia broadcasting). This AV decoder chip comprise all units required for T-DMB multimedia decoding such as system demultiplexer, MPEG-4 AVC/H.264 video decoder and MPEG-4 BSAC audio decoder. The proposed audio/video decoder has low power consumption and has been implemented using a standard-cell library in 0.18 mum 1P6M CMOS technology