To address the waste of spectrum resources caused by the separate operation of integrated sensing and communication (ISAC) and mobile edge computing (MEC), the integrated sensing, communication, and computing (ISCC) paradigm has been proposed. However ISCC systems are still facing serious channel fading and obstacle occlusion problems, which are expected to be solved by the emerging simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS). In this paper, we propose a STAR-RIS-assisted ISCC framework, where the computational tasks from users are offloaded to a base station (BS) for processing with the assistance of a STAR-RIS, and the results are subsequently downloaded back to the corresponding users. Notably, target sensing operation is executed concurrently throughout the process of result downloading. To enhance communication efficiency, the non-orthogonal multiple access (NOMA) scheme is incorporated. The number of offloading bits of users is maximized by optimizing the uplink-downlink resource of the network, including the received beamforming vector, active beamforming vector, time slot, phase shift of STAR-RIS, and computational resource. In order to solve the highly complex non-convex problem, we employ the block coordinate descent (BCD) framework to decompose the proposed problem into three subproblems. The convex-concave procedure (CCCP) method is used to deal with the nonconvex constraints. For the rank-1 constraints, the semidefinite relaxation (SDR) approach and the penalty method are invoked. The numerical results show that: 1) the application of STAR-RIS significantly improves the offloading capability of the network, 2) the NOMA scheme significantly outperforms the orthogonal multiple access (OMA) scheme, and 3) comparing with other benchmark schemes, the proposed algorithm significantly improves the overall network performance.

This article studies the energy efficiency optimization of cargo uncrewed aerial vehicle (UAV) pickup systems, with constraints on on-board energy and load capacity. In the UAV-enabled cargo pickup system, minimizing the total energy consumption and ensuring the safe flight of the cargo UAV is a problem to be solved. However, due to building blockages, the channel between the UAV and ground base stations (GBSs) frequently switches between line-of-sight (LoS) and non-line-of-sight (NLoS), thereby affecting the UAV’s communication quality. This effect is further aggravated by environmental noise interference. Moreover, limited by the on-board energy, it is unrealistic for the UAV to pick up all the cargo in a single flight without charging or replacing the battery. To address the above-mentioned challenges, we propose a UAV pickup system energy efficiency optimization (UPSEEO) framework. In this framework, the UAV’s trajectory between any two pickup points is optimized via the A${}^{*}$algorithm to ensure the stability of the UAV communication link. Next, we employ the particle swarm optimization (PSO) algorithm to optimize both task allocation and flight speed to minimize the total energy consumption, subject to constraints on UAV on-board energy limits and payload capacity. Numerical results show that the proposed framework can ensure the UAV’s communication quality in any spatial topology, with an improvement in energy efficiency of approximately 5% to 50% compared to the comparison experiment.

Reconfigurable intelligent surfaces (RIS) are emerging as a crucial technology to address the energy consumption challenges posed by the widespread deployment of access points (APs) in cell-free massive multiple-input multiple-output (CF mMIMO) systems within future sixth-generation (6G) networks. However, most existing studies on RIS-aided CF mMIMO systems assume ideal hardware and static channel conditions, which deviate from practical deployment scenarios. This work analyzes the performance of a RIS-aided CF mMIMO system by incorporating the combined effects of hardware impairments from non-ideal transceivers and channel aging caused by user mobility. We first characterize both direct and cascaded channels between APs and user equipment, modeling them using correlated Rician fading to capture realistic propagation effects. The overall channel is then estimated via the minimum mean square error method under perfect and imperfect line-of-sight phase knowledge, and we derive an analytical expression for the instantaneous spectral efficiency (SE). We also derive the closed-form expressions of the use-and-then-forget bound with the maximum-ratio transmission precoding method. Building on these insights, we establish an efficient joint optimization framework for beamforming in the AP and phase-shift adaptations in the RIS, exploring an alternating optimization method and a deep-reinforcement learning (DRL)-based algorithm. The numerical results validate our theoretical analysis, illustrating the impact of hardware impairments and channel aging on SE. Although the DRL-based method is scalable and adapts well to dynamic environments, its high computational and memory demands pose challenges for real-time deployment, highlighting a trade-off between performance and feasibility.

With the advancement of integrated circuit (IC), the process technology becomes more complicated and the design margin shrinks. Thus, the parasitic extraction is more demanded during IC design. In this invited paper, we survey the research progress on IC capacitance extraction, especially the usage of deep-learning technologies in relevant problems. Firstly, a method based on graph neural network (GNN) for predicting the parasitic capacitances in the pre-layout design stage is presented. It exhibits potential benefit for the optimization of SRAM design. Then, the deep-learning-inspired methods for post-layout capacitance extraction are presented, including CNN-Cap, NAS-Cap and GNN-Cap, etc. They can revamp the accuracy drawback of layout parasitic extraction (LPE) method and the efficiency drawback of 3-D capacitance field solver. Lastly, we briefly review the deep-learning technique for improving the accuracy of the random walk based 3-D capacitance solver for the structures under the advanced process technology.

This paper explores cargo delivery strategies and path planning for unmanned aerial vehicle (UAV) assigned with kinematically-constrained multi-user pickup and delivery operations. Specifically, the UAV is programmed to depart from a warehouse to complete user-generated orders and then delivery packages to designated user locations. To ensure the safety and efficiency of UAV operations, it is essential that the cargo UAV maintain a reliable connection with ground base stations (GBSs) throughout the entire flight missions. Our aim is to simultaneously enhance both the cargo delivery efficiency and energy efficiency of the system by jointly optimizing the delivery sequence and flight trajectories of the UAV. Initially, we create an urban radio map, which serves as the foundation for implementing a simulated particle swarm optimization (SPSO) algorithm. This approach efficiently determines the optimal sequence of UAV visits to the warehouse and user locations. Subsequently, using these sequences, we apply a hybrid grid search algorithm (HGSA) to meticulously plan the trajectory between pickup and delivery locations according to our objective function. Simulations comparing traditional traveling salesman problem (TSP) models and our proposed time and energy-constrained TSP with kinematic constraints (TETSPKC) show that our approach offers significant optimization advantages.

In this paper, we propose an unmanned aerial vehicle (UAV) assisted integrated sensing and communication (ISAC) system for the Internet of Things (IoT), where the UAV beams simultaneously sense the status information of multiple IoT sensing nodes and transmit the sensing data to a data center. The main objective is to enable the UAV to perform sensing using radar beams and then transmit the collected information to the data center via communication beams. To evaluate the sensing performance of the ISAC system, we introduce radar mutual information as a key metric from an information-theoretic perspective. Considering the mutual interference between sensing and communication as well as the communication rate limitations, we aim to maximize radar mutual information by optimizing the UAV flight trajectory and transmit beamforming. To address this problem, we propose a joint optimization algorithm that employs semi-finite and concave programming techniques to maintain rank-1 constraints with low complexity. Numerical results demonstrate the effectiveness of the proposed algorithm in achieving the maximum radar estimation rate, thereby validating the rationality and feasibility of the proposed design approach.

ABSTRACT This paper investigates a multi‐UAV cargo delivery scenario, where each UAV picks up goods from one location and delivers them to another destination while maintaining connectivity with the ground cellular network. Optimizing task assignment and UAV trajectory design to minimize completion time under the constraints is a significant challenge. To address this, the approach is structured into two principal phases. First, Dijkstra's algorithm is utilized to derive the shortest paths between points while ensuring communication connectivity meets specific quality constraints. Second, these paths are integrated with a novel hybrid optimization algorithm fusing a genetic algorithm and an ant colony algorithm to solve the coupled task assignment and route planning problem subject to communication and payload limitations. The hybrid approach efficiently balances exploration and exploitation, leading to superior task allocation and route planning. Numerical results show that our proposed method is effective in balancing task allocation and reducing overall completion time by comparing it with other integrated optimization techniques.

In this paper, we investigate the trade-off issue between energy efficiency and communication quality in the unmanned aerial vehicle (UAV) enabled cargo delivery system. For a cellular-connected cargo UAV delivering parcels from the warehouse to each user’s location, minimizing both the energy consumption and expected outage time is essential. However, a trade-off exists between these two factors, optimizing one aspect is bound to diminished performance in the other. To jointly reduce the UAV’s energy consumption and expected outage time, we formulate an optimization problem with the objective function to minimize the weighted sum of UAV’s energy consumption and expected outage time. With the aid of radio map, a hybrid deep reinforcement learning (HDRL) algorithm, consisting of an improved ant colony optimization algorithm and the dueling double deep Q network algorithm, is proposed to solve the formulated problem. The delivery sequence and the flight trajectory of the UAV are then jointly optimized by solving the problem with the HDRL algorithm. Numerical results demonstrate that the proposed algorithm effectively reduces both energy consumption and outage time, while achieving a performance improvement of approximately 6% to 50% compared to the comparisons. Moreover, the communication quality of the UAV improves with an increased weight factor, yet gives rise to a higher energy consumption.

In this paper, a cellular-connected uncrewed aerial vehicles (UAVs) pickup and delivery system is studied in which the multiple UAVs are served by ground base station (GBS). These UAVs commence operations from the hangar, systematically execute a sequence of pickups and deliveries before returning, all while maintaining a continuous and reliable communication link with the GBS throughout their missions. Owing to the limited onboard energy resources of UAVs and the influence of task sequence and payload characteristics on UAV’s energy consumption, this study focuses on minimizing energy usage through the optimization of task sequences and flight trajectories. Firstly, a radio map of the operational area is constructed to identify flight zones that ensure reliable communication, an improved Dijkstra algorithm is then proposed to compute the shortest viable path between any two access points that satisfy reliable communication criteria. Furthermore, a chromosome structure tailored for a hybrid genetic algorithm (HGA) is devised to address the complexities of multi-UAV task allocation. With the aid of a developed distance matrix, the HGA is utilized to determine the optimal delivery sequence, thereby achieving the objectives of minimizing the maximum energy consumption (MME) and minimizing the sum energy consumption (MSE) respectively. Finally, simulation analyses demonstrate that the proposed MME and MSE optimization strategies enable approximately a 50% reduction in peak flight energy consumption and around a 40% decrease in total energy consumption, compared to multi-UAV pickup and delivery systems without energy optimization.

In this paper, we consider a cargo autonomous aerial vehicle (AAV)-aided multi-parcel pick-up and delivery network, where the communication ability of the AAV is provided by the ground base stations (GBSs). For such a system setup, our goal is to optimize the trajectory of the cargo AAV while minimizing the combined impact of total energy consumption and total outage time. Simultaneously, we aim to maximize overall user satisfaction throughout the entire flight duration. More specifically, we propose a pick-up and delivery of AAV (PDU) framework to address this problem and this framework consists of two parts. First, a simulated annealing (SA) algorithm is used to obtain the pick-up and delivery (P&D) order of parcels. On the basis of obtaining the P&D order through SA, we further use deep reinforcement learning (DRL) to optimize the flight trajectory of the AAV to ensure the expected communication quality between the AAV and GBSs. To verify the effectiveness of our proposed algorithms, we design three baseline strategies for comparison, and also investigate the effect of using the PDU framework with different weights. Finally, numerical results show that the performance of PDU strategy is improved by about 5%-30% compared with other strategies in solving the performance tradeoff of AAV energy consumption, communication quality, and user satisfaction.

The evolution towards Internet of Things (IoT) in the forthcoming sixth generation (6 G) is facing massive amounts of transmitted data and harsh wireless transmission environment, which severely degrade the quality of communication. To overcome these difficulties, a novel multiple intelligent reflecting surfaces (IRSs) assisted unmanned aerial vehicle (UAV) network framework with non-orthogonal multiple access (NOMA) is proposed in this article, where the UAV applies the NOMA scheme to deliver the information to the ground users assisted by multiple IRSs. We aim to maximize the achievable rate of the considered network while guaranteeing the minimum communication rate of each user, by jointly optimizing the multi-IRS phase shifts, UAV transmit power, UAV trajectory, and NOMA decoding order. To handle the coupled variables and integer constraints, we decompose the original problem into three subproblems based on the block coordinate descent (BCD) framework. Specifically, we first obtain the multi-IRS phase shifts by applying the semidefinite relaxation (SDR) technique. Next, the UAV transmit power allocation is derived by exploiting the concave convex procedure (CCCP) method. The UAV trajectory and NOMA decoding order are finally obtained by invoking the penalty-based method and the successive convex approximation (SCA) technique. Based on these, an alternating optimization algorithm is proposed. The numerical results show that: 1) the NOMA scheme enhances the utilization of the spectrum and enhances the access capacity of the communication system; 2) the multi-IRS cooperative structure increases the reflective channels and effectively improves the air-ground transmission environment, thus enhancing the system achievable rate; 3) the proposed multi-IRS assisted UAV NOMA algorithm achieves a significant network rate improvement compared to other benchmark schemes.

More accurate capacitance extraction is demanded for IC design nowadays. The pattern matching approach and the field solver for capacitance extraction have the drawbacks of in-accuracy and large computational cost, respectively. Recent work [1] proposes a grid-based data representation and a convolutional neural network based capacitance models (called CNN -Cap) for 3- D capacitance extraction. In this work, the techniques of neural architecture search (NAS) is proposed to train better models for 3- D capacitance extraction. Experimental results show that the obtained NAS-Cap model achieves higher accuracy than [1].

Cargo Unmanned Aerial Vehicles (UAVs) have been increasingly utilized in short-distance logistics scenarios within urban environments. This paper delves into the path planning problem for multi-user pickup and delivery operations of UAVs. The UAVs embark from a depot, execute the pickup and delivery tasks, and subsequently return to the depot, all while maintaining a reliable connection with the ground base station (GBS) throughout their flight. The energy consumption of the UAVs is intricately tied to the sequence of pickup and delivery, as well as the weight of the transported goods. To minimize the maximum energy consumption of the UAVs and optimize their flight trajectories, this paper focuses on optimizing the access sequence. Firstly, we construct a radio map of the target area, enabling us to assess communication reliability. Subsequently, we introduce an enhanced Dijkstra’s algorithm to compute the shortest paths between any two access points that ensure reliable communication. Then, we devise a chromosome structure tailored for a hybrid genetic algorithm (HGA), specifically addressing the multi-UAV pickup and delivery problem. Leveraging the constructed distance matrix and the designed chromosome structure, we apply the HGA to solve the problem of minimizing the maximum energy consumption during multi-UAV pickup and delivery operations. Finally, we validate the effectiveness of our proposed method through simulation results.

To achieve higher system energy efficiency, SRAM in SoCs is often customized. The parasitic effects cause notable discrepancies between pre-layout and post-layout circuit simulations, leading to difficulty in converging design parameters and excessive design iterations. Is it possible to well predict the parasitics based on the pre-layout circuit, so as to perform parasitic-aware pre-layout simulation? In this work, we propose a deep-learning-based 2-stage model to accurately predict these parasitics in pre-layout stages. The model combines a Graph Neural Network (GNN) classifier and Multi-Layer Perceptron (MLP) regressors, effectively managing class imbalance of the net parasitics in SRAM circuits. We also employ Focal Loss to mitigate the impact of abundant internal net samples and integrate subcircuit information into the graph to abstract the hierarchical structure of schematics. Experiments on 4 real SRAM designs show that our approach not only surpasses the state-of-the-art model in parasitic prediction by a maximum of 19X reduction of error but also significantly boosts the simulation process by up to 598X speedup.

In this paper, we propose an inspection system based on a cellular-connected unmanned aerial vehicle (UAV), where UAV departs from the starting point and flies to the inspection points for patrol inspection. The purpose of our work is to minimize the total inspection service time of the UAV, while ensuring that the total communication outage time throughout the flight is less than a certain threshold. The total inspection service time encompasses the cumulative time spent by the UAV traveling to each inspection point. To address the non-convex problem, we propose a hybrid algorithm that combines deep reinforcement learning (DRL) with simulated annealing (SA). Initially, we employ DRL to determine the trajectory between any two points within the defined scenario, followed by the utilization of SA to derive the optimal inspection sequence. The numerical results show that the total inspection service time obtained by our proposed algorithm is always lower than the benchmark algorithm, and the effect is better when the threshold is smaller, about 10% lower than the benchmark algorithm.

This paper proposes a joint sensing, communication and computation (JSCC) framework in unmanned aerial vehicle (UAV)-assisted systems, where multi-functional terminal devices (TDs) can perform high-accuracy radar sensing as well as offload computation data to an airborne mobile edge computing (MEC) server over the same frequency band. The key objective of the JSCC framework is to simultaneously minimize the transmitted sensing beampattern matching error whilst maximizing the minimum computation efficiency of TDs. This problem is formulated as a multi-objective optimization problem (MOOP) that jointly optimizes the transmit beampattern, computation offloading, and UAV trajectory. To achieve the computation-sensing trade-off region, we first transform the MOOP into a single-objective optimization problem (SOOP) via the 1-constraint method. To make it more tractable, a generalized Dinkelbach’s and successive convex approximation (GD-SCA) algorithm is proposed. Specifically, GD-SCA transfers the non-convex max-min fractional programming in the resultant SOOP by introducing a general auxiliary polynomial via generalized Dinkelbach’s algorithm. Thereafter, the transmit beampattern, computation offloading, and UAV trajectory optimization are decoupled into two nested subproblems, which can be iteratively solved by invoking successive convex approximation (SCA) method to handle the remaining non-convex components. The proposed GD-SCA can obtain high-quality suboptimal solutions of the original MOOP. We validate the effectiveness of the proposed algorithm by considering two multiple access techniques, i.e., non-orthogonal multiple access (NOMA) and space-division multiple access (SDMA). Simulation results demonstrate that the proposed algorithm can achieve an improved computation-sensing trade-off region compared to conventional schemes especially when exploiting NOMA. Moreover, the multi-functional performance can be significantly improved while stringently guaranteeing both radar sensing and computation offloading requirements.

In this paper, we consider an unmanned aerial vehicle (UAV)-based patrol inspection scenario, where the cellular-connected UAV traverses multiple pre-determined waypoints for data collection, and then offloads computation task to the ground base stations (GBSs). This paper aims to minimize the sum of the total energy consumption by jointly optimizing the task completion time, communication scheduling, computation resource allocation, and UAV's trajectory. First, we decompose the original problem into two trackable subproblems: 1) design the optimal traverse order among cruise points; and 2) determine the optimal transmission strategy between two consecutive cruise points. Then, by involving the communication rate performance and the topology construct among the GBSs and the cruise points, a novel weighted factor of the edge is proposed to design the traverse order, which can be compatible with the light and heavy task offloading scenarios. The successive convex approximation (SCA) technique and block coordinate descent (BCD) framework are adopted to optimize the UAV's trajectory and the wireless resource allocation. The numerical results finally indicate that our proposed transmission strategy solution decreases the total energy consumption in various scenarios and outperforms other benchmark schemes.

Deep neural networks (DNNs) are vulnerable to adversarial examples. And, the adversarial examples have transferability, which means that an adversarial example for a DNN model can fool another model with a non-trivial probability. This gave birth to the transfer-based attack where the adversarial examples generated by a ate model are used to conduct blackbox attacks. There are some work on generating the adversarial examples from a given surrogate model with better transferability. However, training a special surrogate model to generate adversarial examples with better transferability is relatively under-explored. This paper proposes a method for training a surrogate model with dark knowledge to boost the transferability of the adversarial examples generated by the surrogate model. This trained surrogate model is named dark surrogate model (DSM). The proposed method for training a DSM consists of two key components: a teacher model extracting dark knowledge, and the mixing augmentation skill enhancing dark knowledge of training data. We conducted extensive experiments to show that the proposed method can substantially improve the adversarial transferability of surrogate models across different architectures of surrogate models and optimizers for generating adversarial examples, and it can be applied to other scenarios of transfer-based attack that contain dark knowledge, like face verification. Our code is publicly available at https://github.com/ydc123/DarkSurrogate-Model.

Deep neural networks (DNNs) have been shown to be vulnerable to adversarial examples. Moreover, the transferability of the adversarial examples has received broad attention in recent years, which means that adversarial examples crafted by a surrogate model can also attack unknown models. This phenomenon gave birth to the transfer-based adversarial attacks, which aim to improve the transferability of the generated adversarial examples. In this paper, we propose to improve the transferability of adversarial examples in the transfer-based attack via masking unimportant parameters (MUP). The key idea in MUP is to refine the pretrained surrogate models to boost the transfer-based attack. Based on this idea, a Taylor expansion-based metric is used to evaluate the parameter importance score and the unimportant parameters are masked during the generation of adversarial examples. This process is simple, yet can be naturally combined with various existing gradient-based optimizers for generating adversarial examples, thus further improving the transferability of the generated adversarial examples. Extensive experiments are conducted to validate the effectiveness of the proposed MUP-based methods.

In this paper, we consider a multi-user oriented UAV cargo delivery system, cellular-connected UAV fly to all users within the distribution range successively from the starting point to deliver goods. The UAV needs to complete its mission quickly and maintain good communication with ground base stations (GBSs). To satisfied the above requirements, we propose a three-step approach. Firstly, the influence of cargo weight on UAV energy consumption is considered, we propose a weight change travel salesman problem (WCTSP) to desgin initial trajectory. Secondly, the entire flight trajectory is divided into a series of sub-trajectories base on the obtained initial trajectory. Finally, deep reinforcement learning (DRL) is adopted to optimize all the subtrajectives. By setting reasonable neural network parameters and reward function, the optimal trajectory under the current standard can be obtained after the neural network is trained continuously until it converges. This paper aims to minimize the weighted sum of total energy consumption and total outage time by jointly optimizing cargo distribution scheduling, communication scheduling and UAV flight strategy. The simulation results demonstrate the effectiveness of our proposed trajectory optimization scheme.

Accurate capacitance extraction is becoming more important for designing integrated circuits under advanced process technology. The pattern matching-based full-chip extraction methodology delivers fast computational speed but suffers from large error and tedious efforts on building capacitance models of the increasing structure patterns. In this work, we propose an effective method for building convolutional neural network (CNN)-based capacitance models (called CNN-Cap) for two-dimensional (2-D) and three -dimensional (3-D) interconnect structures. With a novel grid-based data representation, the proposed method is able to model 2-D pattern structure and 3-D window structure with a variable number of conductors to largely reduce the number of patterns or increase the accuracy. Based on the ability of ResNet architecture on capturing spatial information and the proposed training skills, the obtained CNN-Cap exhibits much better performance over the multilayer perception neural network-based capacitance model while being more versatile. Extensive experiments on a 55 nm and a 15 nm process technologies have demonstrated that the error of total capacitance produced with 2-D CNN-Cap is always within 1.3%, and the error of produced coupling capacitance is less than 10% in over 99.5% probability. For 3-D structures, CNN-Cap predicts the total capacitance with less than 5% error in 99% probability and with a maximum error of 7.7%. For the tested 2-D and 3-D structures, the CNN-Cap run on a GPU server is more than 4,000× and 12,000×, respectively, faster than the conventional field solver Raphael, while consuming negligible memory.

This paper studies a mobile edge computing (MEC) framework for cellular-connected multiple unmanned aerial vehicles (UAVs), where the UAVs compute the tasks locally or offload them to ground base stations (GBSs). Considering the time-varying characteristics of the task arriving, we formulate a stochastic problem to minimize the system's average weighted sum energy consumption, by optimizing UAV-GBS association, communication and computation resource allocation, and UAVs' trajectories. We apply Lyapunov approach to convert the stochastic problem into a deterministic problem that is then solved by invoking Lagrange duality method and successive convex approximation technique, based on which an online joint optimization algorithm is proposed. Moreover, we design a velocity-triggered penalty term (VTPT) to reduce the UAVs' energy. Numerical results validate the effectiveness of the designed VTPT, and also demonstrate that our proposed algorithm not only decreases the energy consumption but also maintains the task queue stability.

In response to the ever-increasing demands of users for delay-sensitive applications, issues on shortening the task completion time in the mobile edge computing (MEC) system has aroused widespread concern. From the perspective of task execution order, this work provide a task scheduling scheme for multiple edge nodes (EN) while federated learning (FL) is utilized for the collaboration of the ENs in the MEC system. First, to acquire an efficient execution order for the pending computational tasks that dynamically generated on one edge node, a task scheduling algorithm based on deep Q network (DQN) is proposed, which reduces the average task completion time. Then, based on the federated learning, of which characteristic matches the edge system well, we integrate an aggregation mechanism to take advantage of every participating edge node to obtain a set of global parameter that optimizes task completion delay for all nodes in the MEC system. Simulations verify the effectiveness and the superiority of the proposed algorithm in processing delay-sensitive tasks and analyze the key factors that contributes to the system performance.

Mobile-edge computing (MEC) networks are facing limited coverage and harsh wireless transmission environments that severely hinder the computation capacity of the Internet-of-Things (IoT) devices. To overcome these issues, this article proposes a novel MEC framework empowered by an unmanned aerial vehicle (UAV) relay and a reconfigurable intelligence surface (RIS). To fully exploit the potentials in terms of computation enhancement brought by the joint UAV and RIS design, we formulate a max–min computation capacity problem via determining the uplink signal detection, active beamforming of UAV, passive beamforming of RIS, time slot partition, computation bits of UAV, and UAV’s trajectory. We develop a concave–convex procedure (CCCP)-based algorithm in an alternating optimization manner over three subproblems to solve the formulated problem. It finds that the CCCP-based algorithm is conducive to decouple the intractable expressions by converting them into new but tractable second-order cone (SOC) constrains. To evaluate the performance of the proposed CCCP-based algorithm, we later design a direct algorithm by exploiting the implicit convexity of the problem. Simulation results demonstrate that the proposed CCCP-based algorithm derives a comparable performance as the direct algorithm, and achieves about 2.57-Mb max-min computation capacity higher compared with the straight flight case, and 8.08-Mb max–min computation capacity higher compared with the case without RIS, which validate the superiority of the joint UAV and RIS design for computation enhancement.

In this paper, we consider a mobile edge computing (MEC) network where multiple cellular-connected unmanned aerial vehicles (UAVs) can offload their computation tasks to multiple ground base stations (GBSs). In practice, the UAVs are generally unable to master stochastic information of task arrival and channel changes in advance, which may cause a severe issue in terms of energy consumption. Therefore, we formulate a stochastic optimization problem with the goal of minimizing the average weighted sum energy consumption, by jointly optimizing UAV-GBS associations, communication and computation resource allocation, and three-dimensional (3D) UAV trajectories, during which a velocity-triggered penalty term (VTPT) is designed to suppress a large amount of the energy consumption of the UAVs. To handle the stochastic problem, we propose an online resource allocation and trajectory optimization algorithm with outer and inner structures. The outer structure transforms the original problem to a deterministic one by applying the Lyapunov-based optimization framework. The inner structure solves the obtained deterministic problem via the Lagrange duality method and the successive convex approximation technique, based on the block coordinate descent framework. Numerical results demonstrate that: 1) VTPT dramatically decreases the UAVs’ energy consumption, and 2) the proposed algorithm not only reduces the energy consumption but also ensures the computation queue stability compared with other benchmark schemes.

In this work, we present effective quantization approaches for compressing the deep neural networks (DNNs). A key ingredient is a novel dynamic programming (DP) based algorithm to obtain the optimal solution of scalar K-means clustering. Based on the approaches with regularization and quantization function, two weight quantization approaches called DPR and DPQ for compressing normal DNNs are proposed respectively. Experiments show that they produce models with higher inference accuracy than recently proposed counterparts while achieving same or larger compression. They are also extended for compressing robust DNNs, and the relevant experiments show 16X compression of the robust ResNet-18 model with less than 3% accuracy drop on both natural and adversarial examples.

Accurate capacitance extraction is becoming more important for designing integrated circuits under advanced process technology. The pattern matching based full-chip extraction methodology delivers fast computational speed, but suffers from large error, and tedious efforts on building capacitance models of the increasing structure patterns. In this work, we propose an effective method for building convolutional neural network (CNN) based capacitance models (called CNN-Cap) for two-dimensional (2-D) structures in full-chip capacitance extraction. With a novel grid-based data representation, the proposed method is able to model the pattern with a variable number of conductors, so as to largely reduce the number of patterns. Based on the ability of ResNet architecture on capturing spatial information and the proposed training skills, the obtained CNN-Cap exhibits much better performance over the multilayer perception neural network based capacitance model while being more versatile. Extensive experiments on a 55nm and a 15nm process technologies have demonstrated that the error of total capacitance produced with CNN-Cap is always within 1.3% and the error of produced coupling capacitance is less than 10% in over 99.5% probability. CNN-Cap runs more than 4000X faster than 2-D field solver on a GPU server, while it consumes negligible memory compared to the look-up table based capacitance model.

Unmanned aerial vehicle (UAV) has been widely applied in internet-of-things (IoT) scenarios while the security for UAV communications remains a challenging problem due to the broadcast nature of the line-of-sight (LoS) wireless channels. This article investigates the security problems for dual UAV-assisted mobile edge computing (MEC) systems, where one UAV is invoked to help the ground terminal devices (TDs) to compute the offloaded tasks and the other one acts as a jammer to suppress the vicious eavesdroppers. In our framework, minimum secure computing capacity maximization problems are proposed for both the time division multiple access (TDMA) scheme and non-orthogonal multiple access (NOMA) scheme by jointly optimizing the communication resources, computation resources, and UAVs' trajectories. The formulated problems are non-trivial and challenging to be solved due to the highly coupled variables. To tackle these problems, we first transform them into more tractable ones then a block coordinate descent based algorithm and a penalized block coordinate descent based algorithm are proposed to solve the problems for TDMA and NOMA schemes, respectively. Finally, numerical results show that the security computing capacity performance of the systems is enhanced by the proposed algorithms as compared with the benchmarks. Meanwhile, the NOMA scheme is superior to the TDMA scheme for security improvement.

With the high altitude and flexible mobility, unmanned aerial vehicle (UAV) assisted mobile edge computing (MEC) is becoming a promising technology to cope with the computation-intensive and latency-critical task in prospective Internet of Things. In this paper, we propose a novel MEC system with several ground servers at access points and one aerial server carried by UAV. To balance the vital metrics of the MEC system, computation bits and energy consumption, we aim to maximize the weighted computation efficiency of the system, subject to the constraints on communication and computation resources, minimum computation requirement and UAV’s mobility. To this end, a joint optimization problem with the goal of weighted computation efficiency maximization is formulated. First, we analyze the problem and transform it into an equivalent tractable form. Then, we solve the challenging non-convex problem by jointly optimizing the computation task assignment, time slot partition, transmission bandwidth and CPU frequency allocation, transmit power allocation, and UAV’s trajectory, based on the Dinkelbach’s method, Lagrange duality and successive convex approximation technique. Furthermore, we propose an alternative computation efficiency maximization algorithm, followed by the convergence and complexity analysis. Finally, numerical simulations show that our proposed algorithm significantly improves the computation efficiency compared to benchmark schemes. It is also validated that the proposed algorithm effectively obtains a good tradeoff between the computation task bits and energy consumption of the system.

Unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) system is a prominent concept, where a UAV equipped with an MEC server is deployed to serve a number of terminal devices (TDs) of Internet of Things in a finite period. In this article, each TD has a certain latency-critical computation task in each time slot to complete. Three computation strategies can be available to each TD. First, each TD can operate local computing by itself. Second, each TD can partially offload task bits to the UAV for computing. Third, each TD can choose to offload task bits to access point via UAV relaying. We propose a new optimization problem formulation that aims to minimize the total energy consumption including communication-related energy, computation-related energy and UAV's flight energy by optimizing the bits allocation, time slot scheduling, and power allocation as well as UAV trajectory design. As the formulated problem is nonconvex and difficult to find the optimal solution, we propose to solve the problem by two parts, and obtain the near optimal solution by the Lagrangian duality method and successive convex approximation technique, respectively. By analysis, the proposed algorithm can be guaranteed to converge within a dozen of iterations. Finally, numerical results are given to validate the proposed algorithm, which is verified to be efficient and superior to the other benchmark cases.

The standard approach to image instance segmentation is to perform the object detection first, and then segment the object from the detection bounding-box. More recently, deep learning methods like Mask R-CNN perform them jointly. However, little research takes into account the uniqueness of the "human" category, which can be well defined by the pose skeleton. Moreover, the human pose skeleton can be used to better distinguish instances with heavy occlusion than using bounding-boxes. In this paper, we present a brand new pose-based instance segmentation framework for humans which separates instances based on human pose, rather than proposal region detection. We demonstrate that our pose-based framework can achieve better accuracy than the state-of-art detection-based approach on the human instance segmentation problem, and can moreover better handle occlusion. Furthermore, there are few public datasets containing many heavily occluded humans along with comprehensive annotations, which makes this a challenging problem seldom noticed by researchers. Therefore, in this paper we introduce a new benchmark "Occluded Human (OCHuman)", which focuses on occluded humans with comprehensive annotations including bounding-box, human pose and instance masks. This dataset contains 8110 detailed annotated human instances within 4731 images. With an average 0.67 MaxIoU for each person, OCHuman is the most complex and challenging dataset related to human instance segmentation. Through this dataset, we want to emphasize occlusion as a challenging problem for researchers to study.

How can one quickly answer the most and top popular objects at any time, given a large log stream in a system of billions of users? It is equivalent to find the mode and top-frequent elements in a dynamic array corresponding to the log stream. However, most existing work either restrain the dynamic array within a sliding window, or do not take advantages of only one element can be added or removed in a log stream. Therefore, we propose a profiling algorithm, named S-Profile, which is of $O(1)$ time complexity for every updating of the dynamic array, and optimal in terms of computational complexity. With the profiling results, answering the queries on the statistics of dynamic array becomes trivial and fast. With the experiments of various settings of dynamic arrays, our accurate S-Profile algorithm outperforms the well-known methods, showing at least 2X speedup to the heap based approach and 13X or larger speedup to the balanced tree based approach.

In this paper, we consider a ground terminal (GT) to an unmanned aerial vehicle (UAV) wireless communication system where data from GTs are collected by an unmanned aerial vehicle. We propose to use the ground terminal-UAV (G-U) region for the energy consumption model. In particular, to fulfill the data collection task with a minimum energy both of the GTs and UAV, an algorithm that combines optimal trajectory design and resource allocation scheme is proposed which is supposed to solve the optimization problem approximately. We initialize the UAV’s trajectory firstly. Then, the optimal UAV trajectory and GT’s resource allocation are obtained by using the successive convex optimization and Lagrange duality. Moreover, we come up with an efficient algorithm aimed to find an approximate solution by jointly optimizing trajectory and resource allocation. Numerical results show that the proposed solution is efficient. Compared with the benchmark scheme which did not adopt optimizing trajectory, the solution we propose engenders significant performance in energy efficiency.

This paper proposes a new transmission strategy for device‐to‐device (D2D) multicast cooperative communication systems based on Simultaneous Wireless Information and Power Transfer (SWIPT) technology. The transmission block is divided into two slots. In the first slot, the source user transmits the information and energy to the help user by SWIPT. In the second slot, the help user uses the cellular spectrum and forwards the information to multiple receivers by using harvested energy. In this paper, we aim to maximize the total system rate, and to tackle the problem, we propose a two‐step scheme: In the first step, the resource allocation problem is solved by linear programming. In the second step, the power‐splitting coefficient value is obtained by taking the benefit of help user into account. Numerical results show that the proposed strategy not only effectively improves the overall throughput and spectrum efficiency but also motivates the cooperation.

In this study, energy cooperation schemes are considered in wireless cooperative networks; in these networks multiple pairs of users communicate with each other, assisted by an energy harvesting relay that gathers energy from the received signal by applying a power splitting scheme and forwards the received signal by using the harvested energy. This study is focused on the energy cooperation strategies for the relay to distribute the harvested energy between the multiple user pairs in amplify‐and‐forward and in decode‐and‐forward modes. Specifically, optimal solutions without energy cooperation at the relay node are first proposed and then the authors formulate the energy cooperation optimisation problem. This optimisation problem is non‐convex. They propose an iterative energy cooperation solution to maximise the system throughput by assigning the proper harvested energy to each user pair for both types of forwarding models. They show that with the proposed method, the update in each iteration consists of a group of convex problems with a continuous parameter. Moreover, they derive the optimal solution to these convex problems in closed‐form and it is shown that the solution can converge to a local optimum. Simulation results demonstrate that the proposed algorithm outperforms the traditional non‐cooperation method.

In this paper, we study the achievable degrees of freedom (DoF) for an arbitrary MIMO two-way relay channel, where there are K source nodes, each equipped with Miantennas, for i = 1, 2, ⋯, K, and one relay node, equipped with N antennas. Each source node can exchange independent messages with an arbitrary set of other source nodes assisted by the relay. We extend our newly-proposed transmission scheme, generalized signal alignment (GSA) in [1], to the arbitrary MIMO two-way relay channel with antenna configuration satisfying N ≥ Mi+ Mj, ∀i ≠ j. The notion of GSA is to form network-coded symbols by aligning every pair of signals to be exchanged in a projected subspace at the relay. This is realized by jointly designing the precoding matrices at all source nodes and the processing matrix at the relay node. Moreover, the aligned subspaces are orthogonal to each other. Applying the GSA, we show that the DoF upper bound min {Σi=1KMi, 2Σi=2KMi, 2N} is tight under the antenna configuration N ≥ max{Σi=1KMi-Ms-Mt+ds, t| ∀s, t}. Here, ds, tdenotes the DoF of the message exchanged between nodes s and t. In the special case when the arbitrary MIMO two-way relay channel reduces to the K-user MIMO Y channel, we show that our achievable region of DoF upper bound with GSA is larger than the existing result.