6G-driven compute-intensive applications require the collaboration of communication and computing to achieve optimal performance. Such collaboration drives integrated sensing, communication, and computing to support service requirement sensing and on-demand computing task steering within the network. To this end, the Computing-Aware Network proposes to incorporate computing information into the network layer address called service identifier (SID), to implement serviceoriented SID anycast. This integration aims to naturally support dynamic task steering using the anycast mechanism. However, SID represents an abstract service rather than a specific host, making SID anycast incompatible with the TCP communication patterns used by existing socket-based applications. To address this challenge, this paper introduces Srvcast, a host-transparent and stateful anycast solution. Srvcast consists of two stages: WAN routing and edge network forwarding. In WAN routing, it employs a novel service-oriented routing mechanism to ensure connection affinity for anycast. In the edge network forwarding, Srvcast presents ServiceNAT, a P4-based address translation mechanism that enhances SID compatibility with socket-based applications. To implement Srecast, a prototype system is built in a practical WAN environment. The results demonstrate that Srvcast outperforms existing solutions in terms of system complexity and socket connection establishment time. Srvcast can maintain the flexibility of SID anycast while ensuring compatibility with TCP communication patterns at a lower cost.

Practical applications in mobile ad-hoc networks (MANETs) require the support of diverse network services, e.g., host-centric, content-centric, and location-centric routing and forwarding services. However, existing solutions are typically designed over a single network service rather than integrated ones. To embed diverse network services into MANETs, the major challenge is enabling interoperability among various network-layer (L3) protocols without suffering complexity and scalability issues. In this article, we propose OpenL3, a programmable L3 approach to support the coexistence of diverse network services in MANETs. Specifically, OpenL3 first abstracts key attributes from network entities, such as content, locations, or groups of devices. These attributes are embedded into a network address, named multidimensional identifier (MID), to control the routing and forwarding processes. Then, a distributed MID mapping system is established to facilitate efficient MID registration and query. Based on the MID, a programmable routing and forwarding scheme is proposed, which incorporates a lightweight packet processing design using a P4 programmable data plane to enable interoperability among various L3 protocols. A cluster of SDN-based control plane devices collaboratively distribute flow rules to manage data plane behavior. Furthermore, a prototype system is built to implement and evaluate the proposed solutions. Experimental results show that OpenL3 outperforms the existing solutions in terms of end-to-end latency and network throughput while being deployable in MANETs without modifications to network protocols or sockets.

Edge computing (EC), positioned near end devices, holds significant potential for delivering low-latency, energy-efficient, and secure services. This makes it a crucial component of the Internet of Things (IoT). However, the increasing number of IoT devices and emerging services place tremendous pressure on edge servers (ESs). To better handle dynamically arriving heterogeneous tasks, ESs and IoT devices with idle resources can collaborate in processing tasks. Considering the selfishness and heterogeneity of IoT devices and ESs, we propose an incentive-driven multilevel task allocation framework. Specifically, we categorize IoT devices into task IoT devices (TDs), which generate tasks, and auxiliary IoT devices (ADs), which have idle resources. We use a bargaining game to determine the initial offloading decision and the payment fee for each TD, as well as a double auction to incentivize ADs to participate in task processing. Additionally, we develop a priority-based intercell task scheduling algorithm to address the uneven distribution of user tasks across different cells. Finally, we theoretically analyze the performance of the proposed framework. Simulation results demonstrate that our proposed framework outperforms benchmark methods.

With the development of satellite communication technology, satellite-terrestrial integrated networks (STINs), which integrate satellite networks and ground networks, can realize global seamless coverage of communication services. Confronting the intricacies of network dynamics, the resource heterogeneity, and the unpredictability of user mobility, dynamic resource allocation within networks faces formidable challenges. Digital twin (DT), as a new technique, can reflect a physical network to a virtual network to monitor, analyze, and optimize the physical networks. Nevertheless, in the process of constructing a DT model, the deployment location and resource allocation of DTs may adversely affect its performance. Therefore, we propose a STIN model, which alleviates the problem of insufficient single-layer deployment flexibility of the traditional edge network by deploying DTs in multi-layer nodes in a STIN. To address the challenge of deploying DTs in the network, we propose a multi-layer DT deployment problem in the STIN to reduce system delay. Then we adopt a multi-agent reinforcement learning (MARL) scheme to explore the optimal strategy of the DT multi-layer deployment problem. The implemented scheme demonstrates a notable reduction in system delay, as evidenced by simulation outcomes.

Mobile edge computing mitigates the shortcomings of cloud computing caused by unpredictable wide-area network latency and serves as a critical enabling technology for the Industrial Internet of Things (IIoT). Unlike cloud computing, mobile edge networks offer limited and distributed computing resources. As a result, collaborative edge computing emerges as a promising technology that enhances edge networks' service capabilities by integrating computational resources across edge nodes. This paper investigates the task scheduling problem in collaborative edge computing for IIoT, aiming to optimize task processing performance under long-term cost constraints. We propose an online task scheduling algorithm to cope with the spatiotemporal non-uniformity of user request distribution in distributed edge networks. For the spatial non-uniformity of user requests across different factories, we introduce a graph model to guide optimal task scheduling decisions. For the time-varying nature of user request distribution and long-term cost constraints, we apply Lyapunov optimization to decompose the long-term optimization problem into a series of real-time subproblems that do not require prior knowledge of future system states. Given the NP-hard nature of the subproblems, we design a heuristic-based hierarchical optimization approach incorporating enhanced discrete particle swarm and harmonic search algorithms. Finally, an imitation learning-based approach is devised to further accelerate the algorithm's operation, building upon the initial two algorithms. Comprehensive theoretical analysis and experimental evaluation demonstrate the effectiveness of the proposed schemes.

Geo-Distributed Machine Learning (GDML) aims to train large-scale machine learning models across geographically dispersed datacenters. However, the performance of GDML systems is constrained by the limited Wide Area Network (WAN) bandwidth and the presence of the straggler problem. Existing GDML designs often show contradictory effects in addressing these challenges, while in-network computing attempts are typically restricted to single datacenter environments rather than the more complex GDML scenarios. To overcome these limitations, this paper proposes L3DML to facilitate GDML using the P4-based Software-defined Network (SDN). Our approach incorporates three key innovations. Firstly, we introduce a novel network addressing scheme that enables location-specific in-network gradient aggregation for GDML, eliminating the need for parameter servers. Secondly, we utilize the P4 data plane to integrate lossless gradient transmission within switches. Thirdly, we address the straggler problem by employing a unique Deep Reinforcement Learning (DRL) model set and a corresponding rate synchronization routing approach. L3DML is implemented on a prototype system consisting of several Intel Tofino switches and the Spirent network emulator. Experimental results indicate that L3DML outperforms existing solutions in terms of goodput, model accuracy, and training speed gain for large-scale GDML.

Mobile edge computing (MEC) is a promising computing paradigm that offers users proximity and instant computing services for various applications, and it has become an essential component of the Internet of Things (IoT). However, as compute-intensive services continue to emerge and the number of IoT devices explodes, MEC servers are confronted with resource limitations. In this work, we investigate a task-offloading framework for device-assisted edge computing, which allows MEC servers to assign certain tasks to auxiliary IoT devices (ADs) for processing. To facilitate efficient collaboration among task IoT devices (TDs), the MEC server, and ADs, we propose an incentive-driven pricing and task allocation scheme. Initially, the MEC server employs the Vickrey auction mechanism to recruit ADs. Subsequently, based on the Stackelberg game, we analyze the interactions between TDs and the MEC server. Finally, we establish the optimal service pricing and task allocation strategy, guided by the Stackelberg model and priority settings. Simulation results show that the proposed scheme dramatically improves the utility of the MEC server while safeguarding the interests of TDs and ADs, achieving a triple-win scenario.

The potential new application scenarios in the era of 6G will generate a large number of mobile computing tasks. In order to meet the greater resource demands, 6G base stations will not only have traditional communication capabilities but also possess information perception and intelligent computing capabilities, realizing ubiquitous sensing and computing integration. This paper will first introduce the concept and architecture of the 6G computing power dedicated network, which can release the surplus computing power of base stations during idle times in mobile communication services, providing ubiquitous computing services for mobile users near the base station side. The 6G computing power dedicated network can not only reduce the end-to-end latency of services but also effectively improve the utilization rate of edge computing power, avoiding task overload in cloud resource pools.

The rapid expansion of computing needs from emerging applications pushes a large amount of deployment of computing infrastructures and corresponding energy cost and greenhouse gas emissions of computing generate great concern. In this paper, we study how to maximize the platform profit by optimizing task scheduling in the Space-Air-Ground integrated Computing Power Network supplied by green energy while considering both the user requirements and dynamics of green energy. First, we formalize the problem as a binary integer linear programming problem that is NP-hard. The problem is then further modeled as a Markov decision process. Considering the dual dynamics of user requests and the generation of green energy, we propose a task scheduling strategy based on deep reinforcement learning, which can predict power generation based on the current operating status of each hydroelectric power station and also provide a scheduling strategy. Extensive experiments demonstrate that the proposed algorithm performs better than the baseline algorithms.

Low-Earth orbit (LEO) satellite communications play a vital role in global and emergency communications and the movement of LEO satellites around the Earth necessitates frequent handovers for terrestrial users. Efficient handovers are complicated due to the limited coverage and rapid movement of LEO satellites, as well as the high mobility of user equipment (UE), leading to significant handover overhead. To address these challenges, this paper introduces an efficient area-division based handover scheme, in which, the Earth's surface is divided into areas. To reduce the impact of the UE's speed on the handover decision and reduce the handover overhead, handover decisions are made for areas instead of individual UEs based on the satellite trajectories and area characteristics when UEs move within the same area. The graph-based approach is used to compute the handover sequence, which takes multiple handover factors into account to ensure handover performance. We analyse the effects of area size and UE's speed on the overhead and handover failure rate. The experimental results verify that the proposed handover scheme can reduce the handover overhead and improve the handover success rate and data transmission efficiency.

The increasing number of IoT devices in the network brings new challenges to the network carrying capacity of intelligent edge computing, and the complicated network services make the demand for network resources in industrial production scenarios or ordinary network users often exceed the carrying capacity of the edge computing network. To alleviate this problem, this paper proposes an intelligent edge computing architecture that introduces network service identification, extracts and analyses the data characteristics of network traffic, and designs appropriate algorithms to classify network traffic into six different service types. This enables real-time and computing-requiring tasks to be prioritised in the network. Using two machine learning algorithms, KNN and MLP, a model validation is carried out on the constructed dataset, and the results show the effectiveness of the method, with the correct rate of data validation reaching 85%, which is more than 5% higher than the correct rate of direct classification of the specified applications, and the accuracy can be as high as 97% in certain scenarios.

With the development of the Internet of Things (IoT), certain IoT devices have the capability to not only accomplish their own tasks but also simultaneously assist other resource-constrained devices. Therefore, this article considers a device-assisted mobile edge computing system that leverages auxiliary IoT devices to alleviate the computational burden on the edge computing server and enhance the overall system performance. In this study, computationally intensive tasks are decomposed into multiple partitions, and each task partition can be processed in parallel on an IoT device or the edge server. The objective of this research is to develop an efficient online algorithm that addresses the joint optimization of task partitioning (TP) and parallel scheduling (PS) under time-varying system states, posing challenges to conventional numerical optimization methods. To address these challenges, a framework called online task partitioning action and parallel scheduling policy generation (OTPPS) is proposed, which is based on deep reinforcement learning (DRL). Specifically, the framework leverages a deep neural network (DNN) to learn the optimal partitioning action for each task by mapping input states. Furthermore, it is demonstrated that the remaining PS problem exhibits NP-hard complexity when considering a specific TP action. To address this subproblem, a fair and delay-minimized task scheduling (FDMTS) algorithm is designed. Extensive evaluation results demonstrate that OTPPS achieves near-optimal average delay performance and consistently high-fairness levels in various environmental states compared to other baseline schemes.

Due to advantages in security and privacy, blockchain is considered a key enabling technology to support 6G communications. Practical Byzantine fault tolerance (PBFT) and RAFT are seen as the most applicable consensus mechanisms (CMs) in blockchain-enabled wireless networks. However, previous studies on PBFT and RAFT rarely consider the channel performance of the physical layer, such as path loss and channel fading, resulting in research results that are far from real networks. Additionally, 6G communications will widely deploy high-frequency signals, such as terahertz (THz) and millimeter wave (mmWave), while performances of PBFT and RAFT are still unknown when these signals are transmitted in wireless PBFT or RAFT networks. Therefore, it is urgent to study the performance of nonideal wireless PBFT and RAFT networks with THz and mmWave signals, to better make PBFT and RAFT play a role in the 6G era. In this article, we study and compare the performance of THz and mmWave signals in nonideal wireless PBFT and RAFT networks, considering rayleigh fading (RF) and close-in free space (FS) reference distance path loss. Performance is evaluated by five metrics: 1) consensus success rate; 2) latency; 3) throughput; 4) reliability gain; and 5) energy consumption. Meanwhile, we find and derive that there is a maximum distance between two nodes that can make CMs inevitably successful, and it is named the active distance of CMs. The results show that the two consensus networks have a lower consensus success rate, higher delay, lower throughput, and lower energy consumption in mmWave than THz. Compared with the wireless RAFT consensus, wireless PBFT consensus has a lower consensus success rate, higher delay, lower throughput, and higher energy consumption. The research results provide important references for the future transmission of THz and mmWave signals in PBFT and RAFT networks.

With the development of the Internet of Vehicles (IoV), multiaccess edge computing (MEC) technology places computing resources closer to users at edge nodes, enabling faster, more reliable, and secure computing services. In the MEC-enabled IoV networks, task offloading scheduling, as an effective method to alleviate the computational burden on vehicles, is gaining increasing attention. However, with the intelligent and networked development of vehicles, the complex data dependency between in-vehicle tasks brings challenges to offloading scheduling. In contrast to many existing methods that solely address individual tasks, there is a growing need to tackle interrelated tasks within the IoV framework. This includes tasks like processing vehicle sensor data, gathering and analyzing road condition information, facilitating collaborative decision making among vehicles, and optimizing traffic signal systems. Our objective is to address the broader challenge of offloading dependent tasks, as this closely aligns with real-world scenes and requirements. In this article, we propose a priority-based task scheduling algorithm (PBTSA) to minimize processing delay when the tasks are interdependent. PBTSA proposes a method that can better measure the data transmission and calculation delay of the IoV networks. We first model dependent tasks as a directed acyclic graph (DAG) and then use the reverse breadth-first search (RBFS) algorithm to generate the priority of each subtask, and finally according to the priority with low complexity to offload subtasks greedily to minimize task processing delay. We compare the PBTSA with the other two existing algorithms through simulations. The results show that the PBTSA can effectively reduce the task processing delay and can reach close to 10%.

Federated learning effectively addresses issues such as data privacy by collaborating across participating devices to train global models. However, factors such as network topology and computing power of devices can affect its training or communication process in complex network environments. Computing and network convergence (CNC) of sixth-generation (6G) networks, a new network architecture and paradigm with computing-measurable, perceptible, distributable, dispatchable, and manageable capabilities, can effectively support federated learning training and improve its communication efficiency. By guiding the participating devices’ training in federated learning based on business requirements, resource load, network conditions, and computing power of devices, CNC can reach this goal. In this paper, to improve the communication efficiency of federated learning in complex networks, we study the communication efficiency optimization methods of federated learning for CNC of 6G networks that give decisions on the training process for different network conditions and computing power of participating devices. The simulations address two architectures that exist for devices in federated learning and arrange devices to participate in training based on arithmetic power while achieving optimization of communication efficiency in the process of transferring model parameters. The results show that the methods we proposed can cope well with complex network situations, effectively balance the delay distribution of participating devices for local training, improve the communication efficiency during the transfer of model parameters, and improve the resource utilization in the network.

The development of the Internet of Vehicles (IoVs) has attracted much attention due to the increasing number of connected cars. IoV refers to the interconnection of vehicles with other devices through the internet to enable information sharing and interaction. The advent of 5G mobile communication technologies has provided high-speed, low-latency, and high-reliability communication services, which have gone a long way in solving the communication problems associated with IoV. Additionally, the Multi-Access Edge Computing (MEC) technology has placed computing resources on edge nodes closer to the users, thus enabling faster, more reliable, and more secure computing services to meet the vehicles’ computing resource requirements. However, task offloading and resource allocation issues of 5G-connected vehicles enabled by Mobile edge computing remain a significant challenge when it comes to computing tasks and data related to IoVs. Our study proposes a Lyapunov Based Profit Maximum (LBPM) task offloading algorithm, which utilizes the Lyapunov optimization theory to maximize the time-averaged profit as the optimization objective. The algorithm uses the drift plus penalty optimization framework to establish the Lyapunov function and transforms the optimization goal into making a reasonable offloading decision at each time slot to optimize the upper bound of the function. We also compare the LBPM algorithm with existing algorithms for simulation experiments and performance analysis. The experimental results indicate that the LBPM algorithm increases the time-averaged profit by over 15%.

Computing power network (CPN) has been proposed to allocate and schedule computing power resources among cloud, network, and edge according to the needs of computing services. CPN can improve the utilization rate of various computing resource pools. However, it brings other challenges that how to transfer data packets based on computing resource information. Since the size of routing table will be too large to store and search with lots of computing information. To solve this problem, we define three computing service types firstly. Then propose a hierarchical routing mechanism for computing services in CPN. Based on this mechanism, CPN can improve data forwarding efficiency and user experience. In the future, we will research standard of computing resource identification to provide more intelligent service for various application.

Computing power network (CPN) has been proposed to allocate and schedule computing power resources among cloud, network, and edge according to the needs of computing services. CPN can improve the utilization rate of various computing resource pools. However, it brings another challenge that how to ensure the security of multi-tenant information and the resource information which they rent. To solve this problem, we propose an isolation architecture in CPN, add a tenant mapping management module in network control layer firstly. Then we design the security mapping process between the tenant and the computing routing node based on this architecture. At last, we propose a mapping method between tenants and computing routing nodes based on hash ring which can avoid the problem of data migration caused by increasing the number of computing routing nodes. In the future, we will study the mapping algorithm to improve the efficiency of CPN.

Computing power network (CPN), which integrates heterogeneous computing resources and communication network, can tackle the challenges brought by the pervasiveness of mobile and Internet of Things applications. In this paper, we study the optimization of task scheduling in a CPN network by considering the unbalancing between task distribution and resource cost. The design objective is to maximize the system profit while satisfying tasks' delay requirements. We formulate this problem as an integer programming problem. To address this NP-hard problem, we propose a Deep Reinforcement Learning (DRL) based multistage profit-aware task scheduling algorithm which first makes coarse grained task allocation using DRL among regions and then determines an optimized intra-region task assignment by using profit-aware balancing algorithm. Extensive simulations are conducted for performance evaluation and the results show the high performance of the proposed algorithm as compared with baseline algorithms.

Distributed training (DT) has become an important and popular practice for collaborative training of high-quality machine learning (ML) models. The communication efficiency of gradient aggregation has been shown to be the primary performance bottleneck for distributed training today. Advanced programmable switches with in-network computing capabilities provide a promising direction for improving the communication efficiency of DT by offloading some gradient aggregations from the host to switches in the network. In this paper, we propose SPAR to optimize the performance of gradient aggregation under constrained in-network computing capabilities. To improve the aggregation throughput, SPAR jointly optimizes the deployment of in-network aggregation switches and the routing of aggregation requests from workers. We formulate this joint optimization problem as an integer nonlinear programming problem and design an efficient greedy algorithm to compute solutions quickly. The experimental results show that SPAR significantly outper-forms the other state-of-the-art solutions based on in-network aggregation, improving aggregation throughput by up to 3×.

6G will build a Space-Air-Ground Integrated Network (SAGIN) with global wide-area coverage by integrating terrestrial networks (such as mobile cellular networks) and non-terrestrial networks (such as satellite networks and high-altitude communication platforms), thus providing the all-time and all-territory broadband mobile access capabilities for various applications. In this paper, we apply federated learning (FL) to 6G SAGIN network by combining blockchain with it to prevent the adverse effects of malicious devices on models training. As an example, we further illustrate how to utilize our proposed blockchained FL platform to help 6G dynamic spectrum management. Simulation results show that the proposed system offers considerable advantages in terms of detection accuracy and obtained rewards.

Edge computing servers have been widely deployed in recent years to address the requirements of diverse tasks that are sensitive to delays and computationally intensive. However, due to their independent nature and uneven distribution of service requests, certain clusters may be relatively idle, while others may be overloaded. This situation can result in increased latency for certain tasks, and it prevents the full utilization of resources in the edge clusters. To mitigate this problem, we design and implement a prototype testbed for task offloading, aimed at achieving computing and network convergence. This testbed facilitates collaboration among multiple edge computing clusters. We construct multiple clusters using Intel NUC mini computers and incorporate key enabling technologies into the system. We assess the testbed's performance by employing multiple video processing services that require low latency and high computational capacity. In scenarios with uneven service requests, load balancing can be achieved across the edge computing clusters, resulting in reduced response latency for user tasks.

Cloud computing is a unified management and scheduling model of computing resources. To satisfy multiple resource requirements for various application, edge computing has been proposed. One challenge of edge computing is cross-domain data security sharing problem. Ciphertext policy attribute-based encryption (CP-ABE) is an effective way to ensure data security sharing. However, many existing schemes focus on could computing, and do not consider the features of edge computing. In order to address this issue, we propose a cross-domain data security sharing approach for edge computing based on CP-ABE. Besides data user attributes, we also consider access control from edge nodes to user data. Our scheme first calculates public-secret key peer of each edge node based on its attributes, and then uses it to encrypt secret key of data ciphertext to ensure data security. In addition, our scheme can add non-user access control attributes such as time, location, frequency according to the different demands. In this paper we take time as example. Finally, the simulation experiments and analysis exhibit the feasibility and effectiveness of our approach.

Edge computing has been proposed to satisfy delay requirement for various applications. It brings another challenge that the collaboration problem among cloud computing, edge computing and network resources. Diversification of computing power nodes makes the optimization of resource utilization more complicated. To solve this problem, we first describe a general computing power network (CPN) framework and analyze different roles and their focuses on computing power resource. Then we propose a computing power resource modeling approach from the operator's perspective. In this paper, we quantify the computing resource, storage resource and transmission time of wireless and wired. At last, we present completion time of a calculation task to measure computing power for CPN.

The introduction of edge computing has brought new changes in network traffic model. From the perspective of users, resources in different locations are not equal. For providing users with a better service experience, the distance between users and resources (different latency), network conditions, as well as charging and many other factors are needed to be considered. This paper considering various resources proposes an interworking architecture of computing and network based on IP extension: Computing power network. It is a new type of network that realizes the best resource allocation, by distributing computing, storage, algorithm and other resource information of service nodes through network control plane (such as centralized controller, distributed routing protocol, etc.). It combines network context and user requirements to provide the optimal distribution, association, transaction and scheduling of computation storage and network resources.

Software-defined networking (SDN) uses a centralized control plane to manage the whole network. Distributed SDN architecture has been proposed to resolve scalability and reliability problem. One challenge of multiple controllers is load balancing problem, that uneven load distribution among controllers. Dynamic switches migration is an effective way to solve this problem. However, the existing schemes focus on effective migration algorithms and different metrics to realize the load balancing of control plane in SDN, and do not consider the communication cost between controllers and computation cost of migration scheme. In order to address this issue, we propose a load balancing approach based on sharing data store. Sharing data store has the information of all controllers, and the calculator module in it can compute switches migration scheme based saved information. It can also reduce the computation and communication pressure of controllers. Simulation experiments exhibit the feasibility and effectiveness of our approach.