Collaborative inference among Internet-of-Things (IoT) devices can reduce deep neural network (DNN) inference latency by exploiting the communication and computing resources of IoT devices. However, existing collaborative inference strategies often overlook the padding data integrity in information interaction among devices, leading to the loss of boundary data or redundant communication overhead, thus undermining the inference latency improvements. To address these issues, in this paper, we propose PIDC, a padding-aware IoT device collaboration framework for accelerating DNN inference, which jointly optimizes DNN partitioning and padding interaction among devices to minimize inference latency. First, the minimum amount of data exchanged for padding interaction is analyzed. Then, we formulate the latency minimization problem as a nonlinear integer programming problem, and transform it into a linear programming formulation by introducing auxiliary variables, enabling efficient solution with existing solvers. We implement a prototype using heterogeneous devices to validate the effectiveness of PIDC in real-world settings. Experimental results demonstrate that PIDC achieves significant inference latency reductions, with up to 43.0% latency reductions across different DNN models and datasets compared to the state-of-the-art methods.

Task-oriented semantic communication, leveraging learning-based joint source-channel coding (JSCC), has emerged as a key paradigm for low-latency, high-precision edge-assisted Internet of Things systems. However, the direct mapping of source data to continuous channel symbols in JSCC poses a great challenge in compatibility with existing digital systems. To address this, we propose a digital semantic communication scheme, i.e., an AdaptiveSemanticCompression with jointResourceAllocation andModulation (Adaptive-SCRAM) optimization scheme for multi-user image classification. This scheme, with the semantics quantized by a compressed codebook, enables the discrete semantic transmission with adaptive modulation, while achieving high accuracy and low latency with transmission resources optimized in multi-user classification task. Specifically, we first design a vector quantized-variational autoencoder-based digital JSCC framework with regional quantization, by jointly maximizing the semantic entropy and minimizing the codebook training loss with various SNRs and modulation orders considered in Rayleigh fading. Then based on the well trained end-to-end architecture, we mathematically fit the classification accuracy with respect to the effects of both compressed codebook size and received SNR under different modulation orders, providing an effective premise for the task performance optimization. Finally, we consider to maximize the overall multi-user classification accuracy under the transmission delay constraint, by optimizing the compression, modulation, power and bandwidth allocation for each user. To address the highly non-convex issue, we develop a dual-layer optimization algorithm. The outer-layer problem, which optimizes the compressed codebook size and modulation order, is solved by a cross-entropy-based learning algorithm. While for the inner-layer problem, a successive convex approximation method is used to optimize the power and bandwidth allocation. Simulation results show that our JSCC framework significantly reduces the semantic codebook size without compromising the classification accuracy, which is applicable to practical digital transmission systems. More importantly, compared to most existing comparable optimization schemes for image classification, our Adaptive-SCRAM optimization scheme with adaptive compression, modulation, and resource allocation can achieve much higher classification accuracy for multi-user tasks, while guaranteeing the transmission efficiency.

Uncrewed aerial vehicle (UAV) assisted wireless network (UAWN) is emerging as a promising architectural innovation for the provisioning of ubiquitous coverage and enhanced connectivity in the forthcoming 6G era. To accommodate the increasingly diversified services of 6G without deploying individual UAWNs for each service type, the integration of network slicing with UAWNs becomes essential. However, unlike terrestrial networks, the dynamic and uncertain network conditions caused by the mobility of the UAVs pose significant challenges to the UAWN slicing problem. In this paper, we investigate the UAWN slicing problem by jointly considering UAV deployment, channel allocation, and power allocation under uncertain network conditions including imperfect channel state information, uncertain user demand, and imprecise user location. As expected, this problem turns out to be a robust nonconvex mixed-integer problem, making it overwhelmingly difficult to solve. In light of the limited computing power of the UAV, we propose a lightweight optimization named RUNs, which jointly exploits problem decomposition, the augmented Lagrange method, and the batch coordinate descent method. We prove that the RUNs framework runs fast in the sense that it converges to the stationary point at a log-linear rate. Meanwhile, the numerical results demonstrate that RUNs has significant performance gains over existing benchmark solutions.

The data compression based transmission has been envisioned as a promising solution to improve the data transmission efficiency with the limited radio resources in the future sixth-generation (6G) wireless networks. In this paper, we propose an integrated communication and computation resource allocation system for image transmission based on compressive sensing (CS), which consists of several camera devices and a base station (BS). The device side first compresses the images, after which the compressed images are transmitted using non-orthogonal multiple access (NOMA) transmission, and finally the BS restores the received compressed images. Due to the limited energy supply, the total system energy consumption is minimized by jointly optimizing the image sampling rate, the image data transmission power, the number of floating point operations per second (FLOPS), the time of image compression and the time of data transmission under the constraints of latency and the peak signal-to-noise ratio (PSNR). Due to the non-convexity of the proposed problem, after a series of equal substitutions we convexify the problem. Then, the Karush-Kuhn-Tucker (KKT) condition and the gradient descent method are used to obtain the optimal solution of the target problem. After simulation experiments, it is concluded that the proposed CS-based image transmission scheme effectively reduces the total energy consumption by a factor of 2.7 compared with frequency division multiple access (FDMA), and the total latency by 180% compared with the original image transmission.

Due to the limited energy and computing resources of Internet of Things (IoT) devices, the collaboration of IoT devices and edge servers is considered to handle the complex deep neural network (DNN) inference tasks. However, the heterogeneity of IoT devices and the various accuracy requirements of inference tasks make it difficult to deploy all the DNN models in edge servers. Moreover, a large-scale data transmission is engaged in collaborative inference, resulting in an increased demand on spectrum resource and energy consumption. To address these issues, in this paper, we first design an accuracy-aware multi-branch DNN inference model and quantify the relationship between branch selection and inference accuracy. Then, based on the multi-branch DNN model, we aim to minimize the energy consumption of devices by jointly optimizing the selection of DNN branches and partition layers, as well as the computing and communication resources allocation. The proposed problem is a mixed-integer nonlinear programming problem. We propose a hierarchical approach to decompose the problem, and then solve it with a proportional integral derivative based searching algorithm. Experimental results demonstrate our proposed scheme has better inference performance and can reduce the total energy consumption up to 65.3$\%$, compared to other collaboration schemes.

Mobile edge computing (MEC) is emerging as a critical paradigm to meet the growing computational demands of wireless devices. However, edge servers, wireless devices, and service types in MEC networks are usually time-varying due to configurations, traffic patterns, and operational status, which results in inaccurate state estimations. Therefore, existing Deep Reinforcement Learning (DRL)-based offloading algorithms often fail to effectively handle dependency-aware applications. Furthermore, traditional reward functions adopted in DRL-based algorithms fail to decouple historical dependencies among offloading decisions for subtasks, hindering accurate state updates. To address these challenges, we propose a Status Correction-empowered Deep Reinforcement Learning (SC-DRL) algorithm for making the dependency-aware application offloading decisions in this paper. Specifically, we first adopt the State-Adjusted Bellman Equation to ensure accurate updates of DRL state values. Then, we introduce the dynamic estimate equation to enable DRL agents to estimate system states accurately. Furthermore, we mathematically model device load to extend the dynamic estimate equation to handle real-world complexities. Finally, we propose the Reapplying Reward Technology to reduce reward inaccuracy due to historical dependencies. Both simulations and real-world tests show that the SC-DRL improves the ratio of applications completed within their deadlines by an average of 3.36% and 41.94% compared to the state-of-the-art algorithms, such as Advantage Actor-Critic (A2C), Deep Q-Learning (DQN), and Proximal Policy Optimization (PPO).

Unmanned Aerial Vehicles (UAVs) are considered as promising data collection tools because of their maneuverability and line-of-sight conditions, especially for operations at sea. In this paper, we thus deploy a UAV-enabled offshore operating network, in which the UAV acts as an airborne base station and receives data from the sensing devices at sea. Considering the limited spectrum resources, the non-orthogonal multiple access technology is used for data transmission in parallel to improve the spectrum efficiency. In our scheme, we aim to minimize the total system latency by jointly optimizing the trajectory of the UAV and the number of hovering points, under the constraints of the maximum energy threshold of the UAV and the required data size to be collected. Since the proposed problem is non-convex, we use the deep deterministic policy gradient (DDPG) algorithm in the framework of bisection search to obtain the minimum total system latency. Specifically, we first get the optimal UAV trajectory by the DDPG algorithm for a given number of hovering points. Then, the optimal number of hovering points is derived using the bisection search algorithm, based on the requirement of the data amount to be collected. Lastly, the optimal latency is obtained by alternately iterating the DDPG and bisection search algorithms. Through numerical verification, we can effectively minimize the system latency using our proposed algorithm, with a minimum reduction of about 7.4% to a maximum reduction of about 17.7% in comparison with the existing algorithms A2C and DQN.

Internet of Things (IoT) is one of the key applications of 5G, and the data transmission is the basis of IoT networks. In this paper, we investigate the data transmission scheme in non-orthogonal multiple access (NOMA) for IoT networks to minimize the transmission latency. In order to improve the communication efficiency between devices and the base station (BS) without more energy consumption, we deploy an energy harvesting (EH) relay node between devices and the BS for data transmitting and forwarding. Based on this networking model, we first aim at minimizing the transmission latency by jointly optimizing the transmit power of devices and the relay, forwarding ratios among devices, and forwarding time fraction when transmitting a fixed data bits from devices to the BS via the relay under the constraints of energy buffer and data buffer. Noted that the formulated problem is discrete-continuous mixed and non-convex, we apply the deep deterministic policy gradient (DDPG) algorithm in the framework of bisection searching to obtain the optimal solution. Specifically, the bisection searching is used to seek the possible transmission latency, and the DDPG is to check the feasibility of the chosen transmission latency. Finally, the effectiveness of the proposed model-data-driven algorithm is verified by comparing it with other benchmark algorithms, such as LINGO.

Computation offloading via device-to-device communications can improve the performance of mobile edge computing by exploiting the computing resources of user devices. However, most proposed optimization-based computation offloading schemes lack self-adaptive abilities in dynamic environments due to time-varying wireless environment, continuous-discrete mixed actions, and coordination among devices. The conventional reinforcement learning based approaches are not effective for solving an optimal sequential decision problem with continuous-discrete mixed actions. In this paper, we propose a hierarchical deep reinforcement learning (HDRL) framework to solve the joint computation offloading and resource allocation problem. The proposed HDRL framework has a hierarchical actor-critic architecture with a meta critic, multiple basic critics and actors. Specifically, a combination of deep Q-network (DQN) and deep deterministic policy gradient (DDPG) is exploited to cope with the continuous-discrete mixed action spaces. Furthermore, to handle the coordination among devices, the meta critic acts as a DQN to output the joint discrete action of all devices and each basic critic acts as the critic part of DDPG to evaluate the output of the corresponding actor. Simulation results show that the proposed HDRL algorithm can significantly reduce the task computation latency compared with baseline offloading schemes.

Network slicing has been widely recognized as the architectural technology for 5G and beyond wireless network systems to provide tailored service for diverse applications by flexibly splitting and allocating various heterogeneous resources. However, it is still challenging to meet the strict delay requirements of a large number of delay-sensitive applications under traditional slicing architectures. One potential way to tackle this issue is to build network slicing upon Mobile Edge Computing (MEC) systems, where both communication and computing resources are integrated for providing customized service. As such, in this paper, we propose a framework, to jointly optimize communication and computing resources under the scenario of multi-service coexistence, with the objective to minimize the system cost while meeting the diverse QoS requirements. To make the original optimization problem more tractable, we decompose it into two convex sub-problems first. Then we obtain the optimal solutions of the two sub-problems respectively, and finally derive the optimal communication and computing resource allocation scheme based on the optimal solutions of these two sub-problems. Simulation results show that our proposed scheme significantly saves the system cost under various scenarios compared with other benchmarks.

Network slicing is identified as a fundamental architectural technology for future mobile networks since it can logically separate networks into multiple slices and provide tailored quality of service (QoS). However, the introduction of network slicing into radio access networks (RAN) can greatly increase user handover complexity in cellular networks. Specifically, both physical resource constraints on base stations (BSs) and logical connection constraints on network slices (NSs) should be considered when making a handover decision. Moreover, various service types call for an intelligent handover scheme to guarantee the diversified QoS requirements. As such, in this article, a multiagent reinforcement LEarning based Smart handover Scheme, named LESS, is proposed, with the purpose of minimizing handover cost while maintaining user QoS. Due to the large action space introduced by multiple users and the data sparsity caused by user mobility, conventional reinforcement learning algorithms cannot be applied directly. To solve these difficulties, LESS exploits the unique characteristics of slicing in designing two algorithms: 1) LESS-DL, a distributed Q-learning algorithm to make handover decisions with reduced action space but without compromising handover performance; 2) LESS-QVU, a modified Q-value update algorithm which exploits slice traffic similarity to improve the accuracy of Q-value evaluation with limited data. Thus, LESS uses LESS-DL to choose the target BS and NS when a handover occurs, while Q-values are updated by using LESS-QVU. The convergence of LESS is theoretically proved in this article. Simulation results show that LESS can significantly improve network performance. In more detail, the number of handovers, handover cost and outage probability are reduced by around 50%, 65%, and 45%, respectively, when compared with traditional methods.

To address the drastic increase of multimedia traffic dominated by streaming videos, mobile edge computing (MEC) can be exploited to accelerate the development of intelligent caching at mobile network edges to reduce redundant data transmissions and improve content delivery performance. Under the MEC architecture, content providers (CPs) can access MEC servers to deploy popular content items to improve users' quality of experience. Designing an efficient caching policy is crucial for CPs due to the content dynamics, unknown spatial-temporal traffic demands and limited storage capacity. The knowledge of users' preference is important for efficient caching, but is also often unavailable in advance. Machine learning can be used to learn the users' preference based on historical demand information and decide the content items to be cached at the MEC servers. In this paper, we propose a learning based cooperative content caching policy for the MEC architecture, when the users' preference is unknown and only the historical content demands can be observed. We model the cooperative content caching problem as a multi-agent multi-armed bandit problem and propose a multiagent reinforcement learning (MARL)-based algorithm to solve the problem. Simulation experiments are conducted based on the real dataset from MovieLens and the numerical results show that the proposed MARL-based caching policy can significantly improve content cache hit rate and reduce content downloading latency in comparison with other popular caching strategies.

To address the increase of multimedia traffic dominated by streaming videos, user equipment (UE) can collaboratively cache and share contents to alleviate the burden of base stations. Prior work on device-to-device (D2D) caching policies assumes perfect knowledge of the content popularity distribution. Since the content popularity distribution is usually unavailable in advance, a machine learning-based caching strategy that exploits the knowledge of content demand history would be highly promising. Thus, we design D2D caching strategies using multi-agent reinforcement learning in this paper. Specifically, we model the D2D caching problem as a multi-agent multi-armed bandit problem and use Q-learning to learn how to coordinate the caching decisions. The UEs can be independent learners (ILs) if they learn the Q-values of their own actions, and joint action learners (JALs) if they learn the Q-values of their own actions in conjunction with those of the other UEs. As the action space is very vast leading to high computational complexity, a modified combinatorial upper confidence bound algorithm is proposed to reduce the action space for both IL and JAL. The simulation results show that the proposed JAL-based caching scheme outperforms the IL-based caching scheme and other popular caching schemes in terms of average downloading latency and cache hit rate.

To address the vast multimedia traffic volume and requirements of user Quality of Experience (QoE) in the next generation mobile communication system (5G), it is imperative to develop efficient content caching strategy at mobile network edges, which is deemed as a key technique for 5G. Recent advances in edge/cloud computing and machine learning facilitate efficient content caching for 5G, where mobile edge computing (MEC) can be exploited to reduce service latency by equipping computation and storage capacity at the edge network. In this paper, we propose a proactive caching mechanism named Learning based Cooperative Caching (LECC) strategy based on MEC architecture to reduce transmission cost while improving user QoE for future mobile networks. In LECC, we exploit a Transfer Learning (TL)-based approach for estimating content popularity, and then formulate the proactive caching optimization model. As the optimization problem is NP- hard, we resort to a greedy algorithm for solving the cache content placement problem. Performance evaluation reveals that LECC can apparently improve content cache hit rate, decrease content transmission cost in comparison with known existing caching strategies.

In dense and heterogeneous networks, the decoupled uplink#x002F;downlink (UL#x002F;DL) access (DUDA) design has drawn great attentions for improving system performance. Energy efficiency (EE) becomes a major concern for densely deployed heterogeneous cellular networks (HetNets). In this paper, we theoretically analyze the energy efficient sleep strategy for DUDA HetNets. Through using stochastic geometry theory, we first examine the applicability of conventional sleep strategy to DUDA networks and design a new DUDA sleep strategy. We then formulate the energy consumption minimization problem and EE optimization problem, and derive the optimal BS sleep probability. Numerical results reveal that conventional sleep strategy may provide inaccurate guidance for sleep design in DUDA networks, which may lead to excessive sleeps and decrease system EE. Meanwhile our DUDA sleep strategy can effectively reduce network energy consumption. We also find that the dense deployment of small cells may generally increase network EE, but this improvement saturates as the BS density further increases.

To address the explosively growing demand for mobile data services in the 5th generation (5G) mobile communication system, it is important to develop efficient content caching and distribution techniques, aiming at significantly reducing redundant data transmissions and improving content delivery efficiency. In heterogeneous cellular network (HetNet), which has been deemed as a promising architectural technique for 5G, caching some popular content items at femto base-stations (FBSs) and even at user equipment (UE) can be exploited to alleviate the burden of backhaul and to reduce the costly transmissions from the macro base-stations to UEs. In this paper, we develop the optimal cooperative content caching and delivery policy, for which FBSs and UEs are all engaged in local content caching. We formulate the cooperative content caching problem as an integer-linear programming problem, and use hierarchical primal-dual decomposition method to decouple the problem into two level optimization problems, which are solved by using the subgradient method. Furthermore, we design the optimal content delivery policy, which is formulated as an unbalanced assignment problem and solved by using Hungarian algorithm. Numerical results have shown that the proposed cooperative content caching and delivery policy can significantly improve content delivery performance in comparison with existing caching strategies.