In this paper, we study a UAV-Parking-Vehicles assisted Mobile Edge Computing (MEC) network for providing enhanced edge computing services, where a UAV serves to relay ground users’ tasks to parking vehicles having idle computing resources in the vicinity for processing.We formulate the problem of average task delay minimization in this case as a long-term discrete mixed integer programming problem. We transform this problem into two subproblems: Service deployment optimization problem at large time scale and task offloading optimization problem at small time scale. For the former, we define a system utility for measuring the effect of a service deployment profile at parking vehicles, formulate the system utility minimization problem for optimizing the service deployment at vehicles, and propose a Computing resource allocation and Genetic Algorithm based Service deployment Algorithm (CGSA) for obtaining optimized service deployment profiles at parking vehicles on a per cycle basis. For the latter, we propose a Bandwidth allocation, Task offloading, and Transmission scheduling Algorithm (BTTA) for determining optimized task offloading profiles for all users on a per time slot basis. Extensive simulation results show the high performance of our proposed algorithms compared with baseline algorithms.

Most existing GAN inversion methods either achieve accurate reconstruction but lack editability or offer strong editability at the cost of fidelity. Hence, how to balance the distortion-editability trade-off is a significant challenge for GAN inversion. To address this challenge, we introduce a novel spatial-contextual discrepancy information compensation-based GAN-inversion method (SDIC), which consists of a discrepancy information prediction network (DIPN) and a discrepancy information compensation network (DICN). SDIC follows a ``compensate-and-edit'' paradigm and successfully bridges the gap in image details between the original image and the reconstructed/edited image. On the one hand, DIPN encodes the multi-level spatial-contextual information of the original and initial reconstructed images and then predicts a spatial-contextual guided discrepancy map with two hourglass modules. In this way, a reliable discrepancy map that models the contextual relationship and captures fine-grained image details is learned. On the other hand, DICN incorporates the predicted discrepancy information into both the latent code and the GAN generator with different transformations, generating high-quality reconstructed/edited images. This effectively compensates for the loss of image details during GAN inversion. Both quantitative and qualitative experiments demonstrate that our proposed method achieves the excellent distortion-editability trade-off at a fast inference speed for both image inversion and editing tasks. Our code is available at https://github.com/ZzqLKED/SDIC.

Mobile Edge Computing (MEC) can provide low-latency and workload-intensive computing services to user equipments. Elastic scaling, service placement, and task scheduling are key techniques affecting the performance of an MEC system. Elastic scaling is to determine the set of active servers and also the amount of computation resources allocated for each service deployed at a server, service deployment is to determine the set of services/applications to be deployed at each server, and task scheduling is to determine how tasks are assigned among different servers. In this paper, study an MEC system where user demands fluctuate spatially and temporally. Our objective is to minimize the total power consumption and task response time. We accordingly formulate the joint optimization of elastic scaling, service placement, and task scheduling in this case as a Mixed-Integer Nonlinear Programming (MINLP). Due to the hardness of the problem, we propose an efficient joint elastic scaling, service placement, and task scheduling algorithm. Simulation results show that our proposed algorithm can effectively reduce the system cost as compared with baseline algorithms.

Mobile crowdsensing (MCS) has been an effective sensing paradigm by utilizing the smart devices carried by mobile users to complete sensing tasks at different locations. An important problem in MCS is how to achieve effective task assignment in the context of opportunistic sensing, where mobile users are selectively recruited to perform tasks in an opportunistic way. However, most existing work in this aspect suppose there are only one service platform and further the sensing qualities of users are known a priori. In this article, we study the task assignment when there are multiple service platforms and further the sensing qualities of users are unknown a priori. The design objective is to maximize the overall sensing qualities of finished tasks at all platforms. For this purpose, we build a multiplatform cooperation framework and formulate the task quality maximization problem in this case as a 0–1 integer linear programming (ILP) problem. We propose a multiplatform-cooperation-based task assignment mechanism (MCTA). MCTA includes two phases. The first phase establishes stable cooperation relationship among platforms while respecting their respective cooperation willingness, and for this phase, we propose a cross-platform cooperation relationship construction algorithm. The second phase performs effective online task assignment, and for this phase, we propose two online multiarmed bandit (MAB) with sleeping -arms-based user selection algorithms using local and global learning, respectively, based on whether cross-platform user-sensing-quality learning is allowed. We derive the regrets of the proposed algorithms and prove that MCTA has the properties of cooperation stability and computation efficiency. Extensive simulation results show the high performance of our proposed MCTA mechanism as compared with the existing work.

Edge information plays a significant role in instance segmentation. However, many instance segmentation methods directly perform pixel-wise classification via fully convolutional networks, which may ignore object edges. In this paper, we propose a novel Edge Guided Network (EGNet), which exploits edge information to improve the mask accuracy, for instance segmentation. Specifically, we propose an edge branch to extract edge information. Then, we use edge information as guidance and fuse it with mask features, in order to enrich the mask features. Furthermore, we propose a Spatial Attention (SA) module and add it to the backbone of our EGNet, enabling the network to focus more on foreground objects. In addition, we incorporate a Semantic Enhancement (SE) module into the edge branch, aiming to obtain additional global context information. Experimental results on the COCO 2017 dataset show the effectiveness of the proposed EGNet.

Mobile crowdsensing is a new paradigm for data collection by utilizing the mobility of sensor-rich hand-held smart devices. One of the key challenges in mobile crowdsensing is how to effectively assign tasks to mobile users in an online manner. In this paper, we study the online task assignment problem in mobile crowdsensing where each task has specific time window for its sensor data collection. The objective is to maximize the total profit of the platform in whole sensing period. We first model the crowdsensing system and formulate the profit maximization problem under study. To address this problem, we propose two heuristic algorithms, one is bipartite-match-based algorithm (BMA) using Kuhn-Munkres algorithm and the other improves the first by using data offloading for data upload cost reduction, if applicable. We present detailed algorithm design for both algorithms and deduce their computational complexities. Finally, simulation results validate the effectiveness of our proposed algorithms.

Traffic light control is vital for the efficiency of urban transportation. Recently, the increasing of vehicles has brought great challenges to the traffic light control system. However, traditional traffic light controlling methods are inefficient due to the sophistications of traffic dynamics. In this paper, we propose a Graph Attention mechanism based Multi-Agent Reinforcement Learning method (GA-MARL) by extending the Actor-Critic framework to improve the efficiency of cooperation in traffic signal control. The proposed algorithm is based on hard-attention and soft-attention mechanism, which can help agent filter information effectively and calculate the importance of other agents. In addition, we complete our algorithm by adopting the framework of Centralized Training with Decentralized Execution (CTDE) to overcome the challenge of non-stationary non-Markovian environments. Simulation results prove that our proposed method outperforms the representative methods in the literature.

With the rapid development of mobile smart devices, mobile crowdsensing has become an attractive paradigm for sensor data collection. In a mobile crowdsensing system, the platform can publish a set of tasks and then recruit suitable mobile users to accomplish these tasks. In this paper, we study the budget-constrained task assignment problem for mobile crowdsensing. We assume users can choose to take different transportations for task execution, and different choices have different task coverages, travel expenses, and travel time. We model the crowdsensing system and formulate the budget-constrained task assignment problem under study. We prove this problem is NP-hard. To address this problem, we propose a Value/Reward Maximum First heuristic algorithm (VRMF). We present the detailed algorithm design and deduce its computational complexity. Simulation results validate the effectiveness of our proposed algorithm.

Batch forwarding using intra-flow random linear network coding (RLNC) has been used to improve the performance of a wireless network constituent of lossy links. However, existing batch-based forwarding mechanisms in this aspect can lead to a lot of bandwidth waste and thus reduced transmission efficiency. In this paper, we design a Sliding WIndow based Multiple batch forwarding mechanism (SWIM) using RLNC. In SWIM, multiple batches are allowed to be sent out simultaneously in a way that the forwarding process is managed by a sliding window. In SWIM, adaptive rate assignment is used to assign bandwidth resources to different batches based on their decoding states at the destination, in order to make full use of the bandwidth resources. Simulation results show that SWIM can achieve improved throughput performance as compared with existing work.

Recently, research on wireless sensor networks with mobile sinks (mWSN) has attracted a lot of attention. The mobility of such sinks often results in unpredictable changes of network topology, and brings big challenge to the design of efficient routing protocols for such networks. In this paper, we focus on design of an energy efficient distributed routing protocol for mWSNs. For this purpose, we propose a lightweight ring-based bidirectional routing protocol, referred to as BI-LRRP. BI-LRRP does not need location information and it performs ring-based routing on multi-ring based network structure for packet delivery. To reduce the transmission cost and also prolong the network lifetime, BI-LRRP uses bidirectional search for finding a mobile sink before actual packet delivery. Simulations results show that the proposed protocol can achieve high performance as compared with existing work.

Recently, inter-session opportunistic network coding has been considered as a promising technology for improving the performance of a wireless multi-hop network (WMN). However, most existing work in this field did not consider the issue of how the wireless medium is accessed could largely affect the performance of localized network coding. In this paper, we theoretically analyze the throughput improvement obtained by combining network coding and transmission scheduling in a WMN. Then we formulate the optimal throughput problem as a minimum length scheduling problem subject to potential coding opportunities and coding based transmission conflict constraints. We further propose a distributed coding aware transmission scheduling mechanism for WMNs. Simulation results show that our proposed mechanism can remarkably improve the network throughput as compared with existing work.

Network coding has been as a new coding paradigm that can significantly improve the throughput performance of a wireless multi-hop network. However, most previous studies either assume a fixed transmission power or do not take the impact of transmission power/rate to the network coding into consideration. Since in many scenarios, the selection of the transmission power/rate has big impact on network coding gains due to the fact the reception and overhearing probability, spatial reuse are both rely on transmission power/rate. Therefore, how to achieve high network throughput by appropriate selection of transmit power level and its corresponding packet transmit rate, network coding gain (if any) via localized network operations has been a critical issue in distributed multihop wireless networks, or alternatively, what is the best trade-off between the space-time resource (level of spatial reuse and transmission time) and the network coding gain? In this paper, our goal is to achieve the best trade-off between transmission power/rate and coding gains. Aiming at this, we propose a decentralized network coding aware power and rate control mechanism to enable each node to adjust its transmit power and data rate such that the network coding gains and the network throughput is maximized. Simulation results show that the proposed mechanism yields higher performance in network throughput as compared with existing work.

This article proposes a destination-driven on-demand multicast routing protocol (D-ODMRP) to improve the multicast forwarding efficiency in mobile ad hoc networks (MANETs). In D-ODMRP, the path from the multicast source to a multicast destination tends to use those paths passing through another multicast destination. If such multiple paths are available, the one leading to the least extra cost is preferred. This destination-driven strategy is introduced into the on-demand construction process of a multicast forwarding structure in a popular multicast protocol ODMRP. Simulation results show that D-ODMRP can significantly improve the forwarding efficiency as compared with ODMRP. Moreover, the destination-driven strategy can also be introduced into other existing multicast routing protocols for MANETs.

Abstract In wireless sensor networks (WSNs), a mobile sink can help eliminate the hotspot effect in the vicinity of the sink, which can balance the traffic load in the network and thus improve the network performance. Location‐based routing is an effective routing paradigm for supporting sink mobility in WSNs with mobile sinks (mWSNs). To support efficient location‐based routing, scalable location service must be provided to advertise the location information of mobile sinks in an mWSN. In this paper, we propose a new hierarchical location service for supporting location‐based routing in mWSNs. The proposed location service divides an mWSN into a grid structure and exploits the characteristics of static sensors and mobile sinks in selecting location servers. It can build, maintain, and update the grid‐spaced network structureviaa simple hashing function. To reduce the location update cost, a hierarchy structure is built by choosing a subset of location servers in the network to store the location information of mobile sinks. The simulation results show that the proposed location service can significantly reduce the communication overhead caused by sink mobility while maintaining high routing performance, and scales well in terms of network size and sink number. Copyright © 2009 John Wiley & Sons, Ltd.

Efficient and reliable communications is a critical issue in wireless networks with lossy links. In this paper, we propose a neighbor-assisted coding aware opportunistic retransmission mechanism to increase the network throughput. The key idea behind our design is as follows. If a node fails to receive a packet due to link loss, its neighboring node(s) receiving the packet can assist the retransmission of the packet, possibly encoded with other packet(s) via localized network coding, if such retransmission is expected to be beneficial. This can effectively reduce the total number of packet retransmissions at the MAC layer. Simulation results show that our proposed mechanism can significantly increase the network throughput as compared with existing work.

Localized network coding is a promising technique to improve the throughput of wireless multihop networks with multiple concurrent unicast sessions. However, most existing mechanisms in this field perform network coding without considering the maximization of joint coding gain among neighboring nodes. In this paper, we study how to improve network performance by maximizing the area-centric coding gains in wireless networks. To achieve this goal, we design an efficient coding-aware transmission scheduling mechanism. Simulation results show that our mechanism can remarkably improve the network throughput as compared with existing mechanisms.

Network coding has been considered as an effective strategy for improving the performance of wireless mesh networks (WMNs) by encoding multiple packets into a single transmission. Existing work shows that integration of network coding and routing at the network layer can achieve good performance in terms of network throughput and packet delay. In this paper, we propose a rate-adaptive coding-aware multiple path routing mechanism for WMNs. The main design objective is to improve the network performance via traffic splitting for maximizing the coding opportunities in the network. Simulation results are used to verify the effectiveness of our proposed mechanism.

Opportunistic routing and network coding have been considered as effective strategies for improving the throughput of wireless mesh networks (WMN). However, most existing work studied opportunistic routing and network coding separately. This has largely limited the ability of the above strategies from effectively improving the network performance. To achieve improved network throughput, in this paper, we propose a coding-aware opportunistic routing mechanism for WMNs. The design goal is achieved by effectively integrating the above two strategies such that decision on each packet forwarding is made with the awareness of potential coding opportunities. Simulation results show that our proposed mechanism can remarkably improve the network throughput.

Location-based routing has been a critical and efficient routing strategy in large wireless sensor networks (WSN) with mobile sinks. However, the performance of location-based routing highly depends on how position information of mobile sinks are managed and updated. This is typically the task of location service. In this paper, we present the design of a hierarchical location service for WSNs with mobile sinks. The main design objective is to greatly reduce the communication overhead for providing location service while maintaining high routing performance. Detailed simulation results are used to verify the high performance of our designed location service.