Search

Multi-modal brain networks characterize the complex connectivities among different brain regions from structure and function aspects, which have been widely used in the analysis of brain diseases. Although many multi-modal brain network fusion methods have been proposed, most of them are unable to effectively extract the spatio-temporal topological characteristics of brain network while fusing different modalities. In this paper, we develop an adaptive multi-channel graph convolution network (GCN) fusion framework with graph contrast learning, which not only can effectively mine both the complementary and discriminative features of multi-modal brain networks, but also capture the dynamic characteristics and the topological structure of brain networks. Specifically, we first divide ROI-based series signals into multiple overlapping time windows, and construct the dynamic brain network representation based on these windows. Second, we adopt adaptive multi-channel GCN to extract the spatial features of the multi-modal brain networks with contrastive constraints, including multi-modal fusion InfoMax and inter-channel InfoMin. These two constraints are designed to extract the complementary information among modalities and specific information within a single modality. Moreover, two stacked long short-term memory units are utilized to capture the temporal information transferring across time windows. Finally, the extracted spatio-temporal features are fused, and multilayer perceptron (MLP) is used to realize multi-modal brain network prediction. The experiment on the epilepsy dataset shows that the proposed method outperforms several state-of-the-art methods in the diagnosis of brain diseases.

In this paper, we investigate the performance analysis of the buffer-aided energy harvesting device-to-device (D2D) communication system. A joint data and energy resource scheduling problem is proposed considering the data packet and energy packet Poisson arrival but interactive departure. Considering the coupling relationship between data and energy, the joint resource scheduling problem is modeled as a data and energy coupled queuing model. We raise the expressions of carrier-to-interference and noise-ratio (CINR) interval division, which contains the dynamic interferences from cellular users and the other D2D users. To compromise between data delay and energy consumption, we propose a novel adaptive power transfer (APT) strategy that allows more energy consumption to improve data delay. Studying user mobility and APT strategies will lead to complex boundary behavior of the system transition state matrix. By random geometry theory and quasi-birth and death method to obtain steady-state transition probability, the expressions of throughput, delay and packet drop rate for both data queue and energy queue are derived. Simulations are demonstrated to verify the accuracy of the theoretical derivation results.

Proactive content caching at the wireless network edge, such as users and small base stations (SBSs), is an effective way to deal with high mobile traffic. In this paper, based on the user mobility and social relationships, we investigate the optimal caching strategy in device-to-device (D2D) communications underlying heterogeneous networks, where several SBSs are within the coverage of a macrobase station (MBS). Except for SBSs, important users (IUs) hired by an operator also cache files. First, assuming that user preference, i.e., the probability distribution of different file requests of users, are unknown, we cluster users and predict each user preference based on their history file requests by fitting to the Zipf distribution. Second, we derive the closed-form expression of the average system cost by jointly considering the mobility of users and the social relationship between them. With the purpose of minimizing the average system cost, we optimize the SBSs and IU caching strategies. The optimization problem is NP-complete. To solve the problem, we demonstrate that this problem belongs to the minimization of a supermodular function over a partition matroid, and thus, we provide a locally greedy caching algorithm with an approximation ratio of 2 to obtain the sub-optimal solution in polynomial time. Finally, since the operator can reduce the system cost by hiring more IUs, requiring higher payments, we reach a tradeoff by determining the number of IUs. The simulation results show that the proposed caching strategy outperforms the traditional caching strategy, and the suboptimal solution obtained by proposing the greedy algorithm is close to the optimal solution.

This paper proposes a 3D wireless charging cylinder with improved compensation topology to generate a stable rotating magnetic field for charging multiple loads around it. The induced voltage of each receiver is kept constant against load variations when the distance between the transmitter and the receiver is fixed. The 3D charging cylinder has a simple configuration with two perpendicular LCL-compensated transmitter coils driven by separate power inverters sharing a common dc source. The system is fully analyzed with the maximum load limit determined, and design guidance is provided. Both theoretical and experimental results demonstrate that the transmitter currents remain constant against load variations when the total load is within the maximum load limit, and the transfer efficiency is increased with the increase of the total reflected impedance. Compared to the conventional series-series (SS) compensated WPT system, the proposed 3D configuration can maintain the transmitter currents to be constant within the maximum load limit so that multiple loads can be added without affecting each other. It has found that the transmitter currents of the proposed configuration are kept constant at their initial values when four loads with the resistance of 0.5Ω each are randomly placed at 21 cm from the center of the charging cylinder. By contrast, the transmitter currents of the SS compensated WPT system are obviously changed together with the load voltages.

Non-orthogonal multiple access (NOMA) technique, which is known to raise the performance of frequency spectrum constrained wireless communication networks, has obtained wide attention. In this paper, a NOMA-based resource allocation problem in the decode-and-forward relaying downlink network is considered. We aim to maximize the average user transmission rate, subject to the total system power constraint and users' quality of service requirement. A joint user-channel assignment and power allocation problem is formulated, which leads to an NP-hard problem requiring an exhaustive search. To tackle this problem, we adopt a decouple optimization method to separate the joint resource problem into two subproblems, a user-channel assignment problem, and a power allocation problem. After solving the two subproblems, a joint alternating optimization algorithm is proposed with low complexity. The simulation results show that the proposed joint optimal scheme can efficiently improve system performance in the limited channels of multiple users relaying networks.

In this paper, we develop a spatiotemporal mathematical model for analyzing the performance of prioritized data transmissions in the device-to-device (D2D) underlaying cellular network. A dynamic interference model of a D2D user is constructed by exploiting thinned Poisson point process to model the D2D user location with data stored in the buffer. A dynamic priority queuing model is adapted to analyze the performance of multiple types of traffic, in which the priority jump strategy is proposed to provide increased transmission opportunity for low-priority user packets. Then, we employ a two-dimensional Geo/G/1 Markov chain to describe a queue model with priority jump and evaluate it using quasi-birth-and-death process approach. An iterative solution is used to compute the steady-state probability distribution and the expressions of performance metrics are obtained. The simulation results show the validity of the theoretical analysis. Moreover, by comparing the dropping probability of the priority queuing model with and without the jump strategy, the rationality of the introduced model is confirmed.

Dynamic spectrum access (DSA) emerging as an effective way of improving the utilization of the scarce spectrum has attracted great attention in the communication field. A key challenge in DSA is to perform an efficient spectrum sensing and sharing mechanism. In this paper, aiming at achieving a maximal spectrum utilization, we propose a privacy-aware crowdsourced spectrum sensing and multi-user sharing mechanism for DSA. Particularly, in the sensing stage, the advanced mobile crowd sensing is adapted to economically provide sufficient candidate sensing helpers for a sensing requestor. Considering the individual rationality and energy consumption, an incentive mechanism based on both monetary and social motivation is designed to motivate the finial participations of the sensing helpers. Moreover, with the increasing attention to individual privacy, a social network and location-based k-anonymity grouping algorithm are proposed to prevent each helper's privacy being attacked by the malicious requestor or mobile users. Then, for a sensing requestor, aiming at achieving a target detection performance with minimal payment, a truthful reverse auction-based winning group selection algorithm is designed. Furthermore, in the transmitting stage, a realistic scenario is considered where multiple transmitters may discover the same idle spectrum based on sensing helpers' detections and will transmit data simultaneously. Thus, we model this problem as a potential-game-based multi-user transmission mechanism where all the transmitters act as game players and will jointly adjust their transmission powers to maximize the global throughput. Accordingly, we also take advantage of an improved differential evolution algorithm for obtaining a better equilibrium solution in a decentralized way. Both the theoretical analysis and the simulation results prove the rationality and superiority of our proposed algorithms.

The scarcity of spectrum turns the next generation wireless network increasingly heterogeneous. Recently, much attention has been focused on exploiting the unlicensed spectrum to leverage traffic burden from the cellular network. Specified in 3GPP Release 13, long term evolution (LTE) wireless local area networks (WLAN) aggregation (LWA) promises an effective approach for the LTE WLAN interworking and unlicensed spectrum utilization. The LWA could improve the data rate by the licensed and unlicensed carriers aggregation. In contrast to the previous studies, where the LWA is always chosen to boost the peak data rate without considering the user payment, we jointly consider the LWA with Wi-Fi offloading in this paper, aiming to strike the balance between user payment and quality of service (QoS) requirement. We formulate the multi-slot modes selection as a finite-horizon Markov decision problem and propose a Delay-Aware LTE WLAN Aggregation (DLWA) algorithm to obtain the optimal transmission modes strategy. Due to the computational complexity of sequential decision problem, we exhibit the threshold structure of LTE usage and develop a low complexity algorithm based on this structure. Moreover, imperfection in LWA backhaul is considered and analyzed in this paper. Simulation results show the DLWA algorithm and threshold-based DLWA algorithm could guarantee the QoS requirement with lower user payment compared with current LTE WLAN interworking schemes.

This paper proposes a simple and low-cost brightness and color control method for a wireless light emitting diode (LED) system. Since the open circuit voltage of the LED in the secondary side is approximately constant, the average current flowing through the LED reflects the average power received from the proposed wireless power transfer system. The brightness of the LED is adjusted by using a pulse width modulation (PWM) average power control method from the primary side directly. The same inductively coupled channel is utilized to establish a wireless communication channel to control the LED color based on a time-division multiplexing method. A wireless communication protocol is defined, and the worst-case condition for designing the primary PWM modulator identified to avoid the misinterpretation by the secondary demodulator between the power regulation and signal transfer. Experimental results have demonstrated the whole dimming capability of the proposed system with full range duty cycle regulation, and a three digit wireless communication channel developed in this paper can effectively control the selection of eight colors.

A functional brain network has attracted much attention due to its capability of characterizing the functional connectivity patterns of the brain. The existing methods for network construction usually rely on the conventional measures such as Pearson correlation to determine the pairwise similarity between each pair of brain regions, thus ignoring the global structure relationship and the information communication flow among different brain regions. To this end, this paper proposes a directional brain network construction method based on effective distance. Specifically, the effective distance can capture the hidden relevance pattern among all the brain regions and provide a directional functional network reflecting information propagation paths among the functional brain areas simultaneously. When estimating the probability of the direction and the strength of the connectivity between two regions, the structure information characterized by their neighbors is also considered in our method. The functional brain network produced by our method is more flexible in uncovering physiological mechanisms of the brain compared with the conventional undirected network. The experiments on two fMRI data analysis tasks, i.e., disease diagnosis and cognitive state detection, show that our method outperforms the conventional functional brain network approaches, including Pearson correlation, structural equation modeling, and sparse representation-based method.

More mobile devices are now being charged wirelessly due to the convenience, ease of use, and competitive power efficiency of wireless power transfer. However, traditional wireless power transfer systems are not applicable if the mobile devices need to be charged while moving. This paper focused on an omnidirectional wireless power transfer system, which has the potential to transfer power to a movable device. And two different maximum power delivery methods are proposed for applications with intermittently and frequently moving devices. By analyzing the relationship between the input power of the system and the load power, the communication link between the transmitter and the receiver is eliminated. Both methods satisfied the need of dynamic charging for the movable device. The parameter identification method is more suitable for the device which moves intermittently, while the gradient descent method aims at charging the frequently moving devices. Based on the circuit model of the omnidirectional WPT system, the two methods have been mathematically modeled, and the corresponding algorithms for maximum power transfer are proposed. Then, a prototype is implemented to validate the proposed methods. Experimental results have shown that both the methods can be used to achieve maximum output power during the movement of power pickups. At steady-state, a maximum power of 37.08W at 70.39% efficiency has been achieved using the parameter identification method; while 36.48W at 69.90% has been achieved using the gradient descent method.