Stealthy False Data Injection Attack (FDIA) that intentionally modifies measurement data of smart grid meters to bypass the traditional bad data detection module is one of menacing cyber attacks in smart grid. Due to requiring no costly labeling abnormal measurement data, deep neural networks (DNNs) based unsupervised FDIA detection has attracted great attentions. However, the existing schemes have two weaknesses. First, most schemes didn’t take into account the inherent spatial relationships between measurements in the grid. Second, for practical usage, the robustness and generalization of the trained FDIA detection scheme will be influenced by potential noisy measurement data. To address the issues above, based on spatial Graph Neural Network (GNN) architecture, a novel FDIA detection and localization scheme is proposed, named as Recursive Variational Graph Autoencoder (ReVGAE). Specifically, our contributions are following. The VGAE module in our proposed ReVGAE innovatively plays dual roles: data and topology reconstructor, and denoising module. The first role aims to simultaneously reconstruct both nodes’ temporal measurements and topological relationship between nodes. In the second role, the outputs of VGAE as the reconstructor (i.e., the reconstructed temporal measurements) are intentionally used as the artificially noisy samples, and recursively fed into VGAE as input to improve the model’s robustness. Then the residual between the finally reconstructed and the observed measurement data on each node is viewed as anomaly score to judge whether FDIA temporally happens on each node. Thorough experiments on a real grid system demonstrate that the proposed ReVGAE outperforms other VAE and GNN based FDIA anomaly detection schemes.

Accurately forecasting wind speed is crucial for efficiently utilizing the renewable energy, stabilizing the energy system and advancing the progress of the decarbonization of our society. However, due to its inherently temporal volatility and intermittency, accurate wind speed forecasting in a wind farm is challenging. Recently, Large Language Models (LLMs) have demonstrated notable performance in abundant natural language processing and computer vision tasks. However, the conventional LLMs fail to learn the complex spatial and temporal correlations of the wind speed data at multiple turbines in a wind farm, which makes wind speed forecasting cant fully benefit from the significant breakthroughs of LLM. To fill in this gap, we propose a novel spatial-temporal cross-attention LLM framework for wind speed forecasting, namely STCA-LLM, composed of alignment phase, and fine-tuning phase. In detail, our contributions are given as follows. First, the alignment phase aligns the general-purpose LLM with task-specific data, i.e., training the LLM with wind speed data. Second, in the fine-tuning phase, two representation learning modules, i.e., convolution network and graph neural network (GNN) are respectively used to extract temporal features of intra time series in each turbine, and the correlation of inter time-series at multiple turbines in a wind farm. Moreover, the cross-attention module is innovatively proposed to establish the connections between spatial and temporal embeddings. Then, the spatial-temporal representation modules and the aligned LLM are fine-tuned in two-stage way. Finally, thorough experiments on real wind speed dataset demonstrate that our proposed STCA-LLM outperforms state-of-the-art time series forecasting models including Transformer-based models, spatial-temporal GNN-based models, and pertained LLM-based models. Our code is available at: https://github.com/Justinzzcj/STCA-LLM

The accurate power load forecasting is beneficial in reasonably arranging the power supply load, and greatly promoting the safety and smooth operation of smart grid. However, on one hand, high volatility exists in electricity load, especially, in which a few spike loads suddenly occurred has significantly affected the accuracy of short term load forecasting. On the other hand, most existing forecasting schemes follows one-size-fits-all paradigm, i.e., train and utilize a unique model to forecast the whole loads, irrespective of the spike or normal loads. Based on the above observation, we propose a novel short-term load forecasting scheme assisted by cooopeative Multi-task learning (MTL)-based spike occurrence prediction, LFAS, which is composed of two stages. At the first stage, a MTL-based spike occurrence prediction is explicitly proposed to forecast whether the future loads would be spike or not. Then, instead of one-size-fits-all scheme, the second stage intentionally selects the suitable data pre-processing technologies and deep neural network (DNN) prediction models for respectively forecasting spike and normal loads. The experimental results on the Europe electricity load dataset from ENTSO-E Transparency Platform demonstrate that our proposed scheme increases the accuracy of spike occurrence prediction, significantly improves both the forecasting accuracy of spike and normal electricity loads, in comparison with the existing one-size-fits-all load forecasting schemes.

With the explosion of information, personalized recommender plays a vital role in almost all economic platforms. Usually, the recommender exploits user–item (UI) interactive data to learn users’ latent interests, and then correspondingly conducts recommendations. To address the problem of sparse interactions, graph neural networks (GNNs) have been used to efficiently learn the latent representations of users and items, through structurally modeling the inter-relationships among users, items, and their attributes as graphs. However, most of the existing GNN-based methods ignore the issue of the irrelevant attributes, which means that some attributes of an item are irrelevant to a specific user’s preference. Incorporating them into a recommendation scheme may introduce noise and decrease recommendation accuracy. Thus, to address the issue above, this article proposes a novel personalized recommender based on learnable attribute sampling and heterogeneous graph neural network (LASGRec) to improve the recommender’s performance. The work’s contributions are mainly threefold. First, based on the user’s interactive history with items, the heterogeneous user–item–attribute (UIA) graph is constructed, and attributes of the items are sampled with a learnable neural network to alleviate the issue of irrelevant attributes. Second, using the pruned UIA, the heterogeneous GNN model is appropriately used to learn representations of users and items. Novelly, the learnt user’s embedding first aggregates the sampled attributes of items interactive with the user, and then aggregates these items. The learnt embedding of each item incorporates two relationships: the interacted users and its sampled attributes. Finally, comprehensive experiments on multiple real-world datasets demonstrate the superiority of the proposed LASGRec over the state-of-the-art deep neural network (DNN-) and GNN-based recommendation schemes.

In e-commerce platforms, the relevant recommendation is a unique scenario providing related items for a trigger item that users are interested in. However, users' preferences for the similarity and diversity of recommendation results are dynamic and vary under different conditions. Moreover, individual item-level diversity is too coarse-grained since all recommended items are related to the trigger item. Thus, the two main challenges are to learn fine-grained representations of similarity and diversity and capture users' dynamic preferences for them under different conditions. To address these challenges, we propose a novel method called the Dynamic Preference-based and Attribute-aware Network (DPAN) for predicting Click-Through Rate (CTR) in relevant recommendations. Specifically, based on Attribute-aware Activation Values Generation (AAVG), Bi-dimensional Compression-based Re-expression (BCR) is designed to obtain similarity and diversity representations of user interests and item information. Then Shallow and Deep Union-based Fusion (SDUF) is proposed to capture users' dynamic preferences for the diverse degree of recommendation results according to various conditions. DPAN has demonstrated its effectiveness through extensive offline experiments and online A/B testing, resulting in a significant 7.62% improvement in CTR. Currently, DPAN has been successfully deployed on our e-commerce platform serving the primary traffic for relevant recommendations.

Abstract As a distributed learning framework, Federated Learning (FL) allows different local learners/participants to collaboratively train a joint model without exposing their own local data, and offers a feasible solution to legally resolve data islands. However, among them, the data privacy and model security are two challenges. The former means that, if original data are used for trained FL models, various methods can be used to deduce the original data samples, thereby causing the leakage of data. The latter implies that unreliable/malicious participants may affect or destroy the joint FL model, through uploading wrong local model parameters. Therefore, this paper proposes a novel distributed FL training framework, namely LDP‐Fed+, which takes into account differential privacy protection and model security defense. Specifically, firstly, a local perturbation module is added at the local learner side, which perturbs the original data of local learners through feature extraction, binary encoding and decoding, and random response. Then, through using the perturbed data, local neural network model is trained to obtain the network parameters that meet local differential protection, to effectively deal with model inversion attacks. Secondly, a security defense module is added on the server side, which uses the auxiliary model and differential index mechanism to select an appropriate number of local disturbance parameters for aggregation to enhance model security defense and deal with membership inference attacks. The experimental results show that, compared with other federated learning models based on differential privacy, LDP‐Fed+ has stronger robustness for model security and higher accuracy for model training while ensuring strict privacy protection.

Summary The power of artificial intelligence of things (AIoT) stems from adapting machine learning (ML) and artificial intelligence (AI) models into abundant intelligent IoT fields, based on a large data stream with different formats, sizes, and timestamps generated by massive numbers of heterogeneous sensors. On the one hand, data acquisition is the fundamental basis for any AIoT systems, but data sensed by massive IoT devices may be noisy and even contain adversarial samples. On the other hand, ensuring the efficiency and robustness in data acquisition is vitally important for data‐driven ML and AI. Recently, besides perceiving ability, the literature has witnessed great development of empowering things with learning and reasoning ability through deep learning models, including recurrent neural networks (RNNs) and/or convolutional neural network (CNNs). However, the existing works have one significant weakness: fail to explicitly leverage the geospatial implications and latent connections among sensors for high‐quality data acquisition and quality control. Graphs are intrinsically suitable for representing the dependencies and inter‐relationships between AIoT data sensing devices. Due to the ability of capturing the complex interactive relationships between nodes and producing high‐level representations of the graph input, graph neural networks (GNNs) have exploded onto various ML and AI fields, to learn from graph‐structured data. Our review covers the latest progresses in GNN for the fundamental atomic task of data acquisition in AIoT. Instead of surveying the abundant GNN schemes in vertically various IoT sensing applications, this paper systematically reviews the horizontal infrastructure that all AIoT fields should have, that is, AIoT data acquisition, based on GNN and other related emerging AI factors. Our contributions include the following aspects: Provide the latest progresses in GNN for the horizontal task of data acquisition in AIoT, propose the unified GNN pipeline based on encoder–decoder paradigm, and systematically categorize and summarize the emerging technologies helpful to address the issues in AIoT data acquisition, especially the noisy and adversarial data, and point out some future directions about GNN‐based AIoT data acquisition.

The smart grid is now increasingly dependent on smart devices to operate, which leaves space for cyber attacks. Especially, the intentionally designed false data injection attack (FDIA) can successfully bypass the traditional measurement residual-based bad data detection scheme. Considering that the smart grid data naturally contain linear and nonlinear components, inspired by parallel ensemble learning, especially by the stacking method, this article presents an effective two-level learner-based FDIA detection scheme using the Kalman filter and recurrent neural network (KFRNN). The first level includes two base learners, in which the Kalman filter is used for state prediction to fit linear data, and the recurrent neural network is used to fit the nonlinear data feature. The second-level learner uses the fully connected layer and backpropagation (BP) module to adaptively combine the results of two base learners. Then, through fitting Weibull distribution of the sum of square errors (SSEs) between the observed measurements and the predicted measurements, the dynamic threshold is obtained to judge whether FDIA occurs or not. Comprehensive simulation results show that our scheme has better performance than other neural network-based and ensemble learning-based FDIA detection schemes.

Couplet is a unique art form in Chinese traditional culture. The development of deep neural network (DNN) technology makes it possible for computers to automatically generate couplets. Especially, Transformer is a DNN-based “Encoder–Decoder” framework, and widely used in natural language processing (NLP). However, the existed Transformer mechanism cannot fully exploit the essential linguistic knowledge in Chinese, including the special format and requirements of Chinese couplets. Therefore, this article adapts the Transformer mechanism to generate meaningful Chinese couplets. Specifically, the contributions of our work are threefold. First, considering the fact that the words in the corresponding positions of the antecedent clause and the subsequent clause in a Chinese couplet always have same part-of-speech (pos, i.e., word class), pos information is intentionally added into the Transformer to improve the accuracy of the conceived couplet. Second, to deal with the large number of unregistered and low-frequency words in Chinese couplet, a specific unregistered/low-frequency word processing mechanism (UWP) is designed and combined with the Transformer model. Third, to further improve the coherence of couplets, we incorporate the polish mechanisms (PMs) into Transformer model. In terms of three evaluation criteria including bilingual evaluation understudy (BLEU), perplexity, and human evaluation, the experimental results demonstrate the effectiveness of our designed Chinese couplet generation system.

Recently, on one hand, human activity recognition (HAR) has witnessed great application on portable smart devices (e.g., smart phones and wearables, etc.) as they are widely used around the world. On the other hand, HAR methods based on deep learning have attracted much attention, for they possess excellent performance due to their strength on extracting virtual features automatically and hierarchically. However, to establish a personalized deep learning based HAR scheme based on smart devices, insufficient records from target users and heavy computation cost on training from scratch are two challenges. Considering that, in transfer learning, the knowledge learnt in the source domain could be appropriately transferred to help accomplish tasks in the target domain, this paper proposes a personalized HAR scheme through exploiting virtual feature adaptation based on transfer learning (i.e., VFAT) to achieve high recognition accuracy with low computation time. VFAT is composed of pre-training phase on sufficient labeled records in source-domain, and adaption phase on target-domain that uses the few labeled records available. Specifically, VFAT scheme pre-trains the LSTM-based feature extraction component in the pre-training phase and then introduces domain loss in the adaptation phase to minimize the similarity between target-domain virtual features and source-domain activity patterns (i.e., virtual features averaged by activity labels). The HAR scheme applied to the MotionSense dataset and results demonstrate the effectiveness of our proposed VFAT scheme. Moreover, we also investigate the impact of domain division on the performance of transfer learning based HAR.

Feature selection (FS) is one of the fundamental data processing techniques in various machine learning algorithms, especially for classification of healthcare data. However, it is a challenging issue due to the large search space. Binary Particle Swarm Optimization (BPSO) is an efficient evolutionary computation technique, and has been widely used in FS. In this paper, we proposed a Confidence-based and Cost-effective feature selection (CCFS) method using BPSO to improve the performance of healthcare data classification. Specifically, first, CCFS improves search effectiveness by developing a new updating mechanism that designs the feature confidence to explicitly take into account the fine-grained impact of each dimension in the particle on the classification performance. The feature confidence is composed of two measurements: the correlation between feature and categories, and historically selected frequency of each feature. Second, considering the fact that the acquisition costs of different features are naturally different, especially for medical data, and should be fully taken into account in practical applications, besides the classification performance, the feature cost and the feature reduction ratio are comprehensively incorporated into the design of fitness function. The proposed method has been verified in various UCI public datasets and compared with various benchmark schemes. The thoroughly experimental results show the effectiveness of the proposed method, in terms of accuracy and feature selection cost.

The crowdsourcing system is a distributed problem-solving platform, in which tasks are delivered to the crowd (i.e., crowdworkers) in the form of an open call. Usually, large-scale crowdsourcing systems contain abundant microtasks, and the overhead of a crowdworker spending on searching the appropriate task may be comparable to the cost of completing the task. Therefore, task recommendation is necessary. However, existing work ignores the dynamics in crowdsourcing system, i.e., new tasks continually arrive, which leads to the issues of task cold-start. To overcome the challenge of the new coming task recommendation, this article proposes a feature-based Bayesian task recommendation (BTR) scheme. The key idea to deal with the dynamics of the crowdsourcing system lies in that the BTR learns the latent factor of the task through the task features instead of task ID and then learns the user's preference according to their historical behaviors. Specifically, based on task features and the user's historical behavior records, BTR can not only timely provide crowdworkers with personalized task recommendations but also solve the task cold-start problem. The simulations based on the real crowdsourced data set demonstrate that BTR performs better than other typical schemes that target at recommending the newly arrived tasks to crowdworkers.

Deep neural network (DNN) learning has witnessed significant applications in various fields, especially for prediction and classification. Frequently, the data used for training are provided by crowdsourcing workers, and the training process may violate their privacy. A qualified prediction model should protect the data privacy in training and classification/prediction phases. To address this issue, we develop a differential privacy (DP)-enabled DNN learning framework, DNN-DP, that intentionally injects noise to the affine transformation of the input data features and provides DP protection for the crowdsourced sensitive training data. Specifically, we correspondingly estimate the importance of each feature related to target categories and follow the principle that less noise is injected into the more important feature to ensure the data utility of the model. Moreover, we design an adaptive coefficient for the added noise to accommodate the heterogeneous feature value ranges. Theoretical analysis proves that DNN-DP preserves ε-differentially private in the computation. Moreover, the simulation based on the US Census data set demonstrates the superiority of our method in predictive accuracy compared with other existing privacy-aware machine learning methods.

Recently, in this article, due to pervasive usability, smartphone-based human activity recognition (HAR) has witnessed significant development in smart health. Meanwhile, deep recurrent neural network (DRNN) shows a strong ability to automatically extract features from the time-series data, and therefore, DRNN-based HAR schemes have achieved more effective recognition (i.e., recognition accuracy) than those adopting the traditional machine learning. However, the efficiency in training and recognition (in terms of running time) has not fully taken into account, especially for resource-constraint smartphones. To solve the above issue, we propose the PSDRNN and tri-PSDRNN schemes that employ the explicit feature extraction before DRNN. Specifically, considering that the power spectral density (PSD) feature can capture the frequency characteristics and meanwhile retain the successive time characteristics of data gathered from smartphone accelerometer, PSD feature vectors are, respectively, extracted from linear accelerations and triaxle accelerations and explicitly used as the input to the following DRNN classification model. Thorough experiments based on a real dataset demonstrate that the PSDRNN can achieve the comparable effectiveness as the xyz-DRNN (the most accurate DRNN-based HAR scheme only using acceleration data), and the average recognition and training time were reduced by 56% and 80%, respectively. Moreover, tri-PSDRNN advantages over the xyz-DRNN in terms of recognition accuracy, and the running time is still lower than the xyz-DRNN. Besides, our proposed PSDRNN scheme achieved superiority in the recognition of complex transition activities.

Crowdsourcing has proved to be a splendid tool to aggregate the knowledge from a pool of individuals in order to perform abundant microtasks efficiently. Recently, with the explosive growth of online social network, Word of Mouth (WoM)-based crowdsourcing systems have emerged, in which besides conducting the tasks by themselves, participants simultaneously recruit other individuals through exploiting their social networks to help solve crowdsourced tasks. This crowdsourcing paradigm can greatly facilitate to grow the pool of crowdworkers. However, there exist two conflicting challenges in designing an effective WoM-based incentive mechanism: 1) sybil attack and 2) heterogeneous effect of participants. That is, intuitively, incentivizing (usually compensating for) common-ability individuals will inevitably stimulate the behavior of sybil attack (i.e., some individuals create multiple sybils, and split the total efforts into those sybils to expect more compensation). This paper proposes a novel biased contest-based crowdsourcing incentive mechanism through exploiting social networks (BciNet), aiming to balance those two conflicting objectives. BciNet is composed of two phases. First, based on spreading activation model, an enhanced geometric virtual point dissemination mechanism is able to provide sybil-proof property and accommodate the realistic social network structure. Second, based on participants' virtual points, a biased contest gives more reward to less able participants. Through carefully calibrating the bias factor, simulation results based on the real dataset show that BciNet can greatly improve the amount of participants' effort levels, and actually be robust against the sybil attack. In brief, for a practical incentive mechanism, the methodology to address conflicting goals is to put rational individuals into dilemma: to be sybil or not to be, it is the problem, i.e., the potential gain from the sybils in the second phase may be offset by the loss in the first phase.

With the rapid development of Internet of Things (IOT) technologies, modern power systems have become complex cyber-physical systems. A large number of smart devices have promoted efficient generation, transmission and distribution in the smart grid. State estimation (SE) is one of fundamental components in smart grid that evaluates the operation state of a grid by using a set of sensor measurements and grid topologies. A major issue is the authenticity of the measurements collected by the sensors. Specifically, the false data injection attack (FDIA) aims to temper the information that reflect the grid operation state. In this paper, we propose an efficient prediction-based FDIA detection and location scheme, PDL, in which the state vector of smart grid can be represented as multivariate time series, and can be predicted by vector autoregressive processes (VAR) through intentionally exploiting the temporal and spatial correlations of states. Different from most previous works which assumed the state transfer matrix constant and diagonal, a time-varying and non-diagonal matrix is adopted in this scheme. Then, the consistency between the predicted measurements and the observed measurements is utilized to detect and locate abnormal data. Besides, the detected abnormal data can be replaced with the predicted data, which simplifies the calibration process. Extensive simulation results verify the performance of the proposed scheme.

Summary Crowdsourcing is the generalized act of outsourcing tasks, traditionally performed by an employee or contractor, to a large group of Internet population through an open call. With the great development of smartphones with rich built‐in sensors and multiple ratio interfaces, mixing smartphone‐based mobile technologies and crowdsourcing offers significant flexibilities and leads to a new paradigm called mobile crowdsourcing (MCS), which can be fully explored for real‐time and location‐sensitive crowdsourced tasks. In this paper, we present a taxonomy for the MCS applications, which are explicitly divided as using human as sensors, and exploiting the wisdom of crowd (i.e., human intelligence). Moreover, two paradigms for mobilizing users in MCS are outlined: direct mode and word of mouth mode. A comprehensive MCS framework and typical workflow of MCS applications are proposed, which consist of nine functional modules, pertaining to three stakeholders in MCS: crowdsourcer, crowdworkers, and crowdsourcing platform. Then, we elaborate the MCS challenges including task management, incentives, security and privacy, and quality control, and summarize the corresponding solutions. Especially, from the viewpoints of various stakeholders, we propose the desired properties that an ideal MCS system should satisfy. The primary goal of this paper is to comprehensively classify and provide a summary on MCS framework, challenges, and possible solutions to highlight the MCS related research topics and facilitate to develop and deploy interesting MCS applications. Copyright © 2016 John Wiley & Sons, Ltd.

This paper focuses on incentivizing cooperative behavior in community-based autonomous networking environments (like mobile social networks, etc.), in which through dynamically forming virtual and/or physical communities, users voluntarily participate in and contribute resources (or provide services) to the community while consuming. Specifically, we proposed a simple but effective EGT (Evolutionary Game Theory)-based mechanism, VPEF (Voluntary Principle and round-based Entry Fee), to drive the networking environment into cooperative. VPEF builds incentive mechanism as two simple system rules: The first is VP meaning that all behaviors are voluntarily conducted by users: Users voluntarily participate (after paying round-based entry fee), voluntarily contribute resource, and voluntarily punish other defectors (incurring extra cost to those so-called punishers); The second is EF meaning that an arbitrarily small round-based entry fee is set for each user who wants to participate in the community. We presented a generic analytical framework of evolutionary dynamics to model VPEF scheme, and theoretically proved that VPEF scheme's efficiency loss defined as the ratio of system time, in which no users will provide resource, is $4/(8+M)$. $M$ is the number of users in community-based collaborative system. Finally, the simulated results using content availability as an example verified our theoretical analysis.

Unlike traditional epidemic virus spreading, behavior diffusion in social networks is conducted by rational users, who would make strategic choices instead of being randomly infected with some probability. Specifically, individuals always try to maximize their utilities through rationally selecting specific behaviors (adopting a new product, or spreading a rumor, etc.). However, utility obtained by an individual, naturally contains uncertainty, and it may stem from two sources: users' imperfect and incomplete knowledge about others, and the inherently stochastic property in human behavior. Thus, it is imperative to model and analyze the diffusion pattern under the resulting uncertainty in social networks, which, however, has not yet been deeply examined by the existing works. This paper deeply explores the pattern of gossip diffusion in social networks when uncertainty exists in users' decision making. In detail, the innovative results provided in this paper are: first, inspired by random utility theory, we formulate the diffusion model based on mixed logit model that allows for user's uncertainty in determining whether to adopt a specific strategy; second, the formal analysis framework characterizing the diffusion process is derived through the approximation method of mean field theory; finally, we explore the extensive applicability of our proposed analysis framework through modeling rumor diffusion in social networks as a coordination game. Our findings are, for various structural characteristics, small uncertainty can significantly speed up the diffusion of gossip; furthermore, social networks with scale-free property can facilitate the gossip diffusion in the easiest way, but, the range of uncertainty factor that can maximize gossip diffusion is the smallest. The obtained results perfectly comply with the philosophical saying about rumor diffusion in real social life: easy come, easy go.

Nowadays, most popular social networking services adopt centralized architecture, in which continual Internet connectivity is prerequisite for each user to exploit those services, and centralized servers are used for storage and processing of all application/context data, even though mobile users are within proximity area (like campus, event spot, and community), and can directly exchange media through various wireless technologies (e.g., Bluetooth, Wi-Fi Direct, etc.). On one hand, the omniscient centralized server may cause serious privacy concern, due to the fact that it collects and stores all users' data (messages, profiles, location, relations, etc.), On the other hand, transmitting a large amount of media generated by users in proximity through Internet, would not only bring a lot of pressure to the network infrastructure and service provider, but incur heavy data traffic cost to users. In this paper, our contributions are twofold. First, we propose a Wi-Fi Direct based P2P social networking framework, which enables direct data exchange among users without using infrastructure network when users are located in proximity, and provide solutions to two core problems in this framework, i.e., discoverability and privacy. Second, the prototype of this framework is preliminarily implemented in Android, which is composed of the following functions: localization based on Google Geocoding, Chat, and file-sharing component supporting intermittent transmission.

Currently, on one hand, there exist much work about network formation and/or growth models, and on the other hand, cooperative strategy evolutions are extensively investigated in biological, economic, and social systems. Generally, overlay networks are heterogeneous, dynamic, and distributed environments managed by multiple administrative authorities, shared by users with different and competing interests, or even autonomously provided by independent and rational users. Thus, the structure of a whole overlay network and the peers' rational strategies are ever coevolving. However, there are very few approaches that theoretically investigate the coevolution between network structure and individual rational behaviors. The main motivation of our article lies in that: Unlike existing work which empirically illustrates the interaction between rational strategies and network structure (through simulations), based on EGT (Evolutionary Game Theory), we not only infer a condition that could favor the cooperative strategy over defect strategy, but also theoretically characterizes the structural properties of the formed network. Specifically, our contributions are twofold. First, we strictly derive the critical benefit-to-cost ratio (b/c) that would facilitate the evolution of cooperation. The critical ratio depends on the network structure (the number of peers in system and the average degree of each peer), and the evolutionary rule (the strategy and linking mutation probabilities). Then, according to the evolutionary rules, we formally derive the structural properties of the formed network in full cooperative state. Especially, the degree distribution is compatible with the power-law, and the exponent is (4-3v)/(1-3v), wherevis peer's linking mutation probability. Furthermore, we show that, without being harmful to cooperation evolution, a slight change of the evolutionary rule will evolve the network into a small-world structure (high global efficiency and average clustering coefficient), with the same power-law degree distribution as in the original evolution model.

Heterogeneities (heterogeneous characteristics) are intrinsic in dynamic and autonomous networks, and may be caused by the following factors: finite nodes, structured network graph, mutation of node's strategy and topological view, and dynamic linking, and so on. However, few works systematically investigate the effect of the intrinsic heterogeneities on the evolutionary dynamics of incentive mechanisms in autonomous networks. In this article, we thoroughly discuss this interesting problem. Specifically, this article respectively models the pairwise interaction between peers as PD (prisoner's dilemma)-like game and multiple peers' interactions as public-goods game, proposes a general analytical framework for dynamics in evolutionary game theory (EGT)-based incentive mechanisms, and draws the following conclusions. First, for explicit incentive mechanisms, due to heterogeneity, it is impossible to get the static equilibrium of absolutely-full-cooperation (or state that provides service to the networks—so-called reciprocation), but, on the other hand, heterogeneity can facilitate reciprocation evolution, and drive the whole system into almost-full-reciprocation state, that is, most of the system time would be occupied by the full reciprocation state. Second, even without any explicit incentive mechanisms, simultaneous coevolution between dynamic linking and peers' rational strategies can not only facilitate the cooperation evolution, but drive the network structure into the desirable small-world structure. The philosophical implication of our work is that simplicity and homogeneity are too idealized for incentive mechanisms in autonomous networks—diversity and heterogeneity are intrinsic for any incentive mechanism that is compatible with the essence of our real society. Diversity is everywhere.

This paper thoroughly investigated the Evolutionary Game Theory (EGT) based modeling and analysis of reciprocation-based incentive mechanisms. Unlike existing work which adopts replicator equation to analyze the stability of incentive mechanisms (actually, replicator equation is only applicable to describe deterministic selection in infinitely large and well-mixed population), we paid special attentions to the intrinsic heterogeneity in real autonomous networks: finite users, mutation probability and structured network graph, and proposed the unified framework to characterize the evolutionary dynamics. Specifically, through modeling and analyzing Prisoner's Dilemma (PD)-like game based and Public-goods game based incentive mechanisms, we show that although it is impossible for incentive mechanisms to get the whole network into static "absolute full cooperation (or reciprocation)" state, they can still drive the whole system into "almost reciprocation" state, that is, most of the system time would be occupied by the cooperation (or reciprocation) state.

In this paper, we propose the DoubleFace algorithm to infer peers’ reputation rankings, which explicitly includes two phases. First is the computation of peers’ Recommendation Ability (RA) based on the following intuitive idea: Peer's RA will be reversely determined by how many peers it points to and how bad those peers have been rated, which comprise two subroutines: the badness propagation and the conversion of RA from badness. In the first subroutine, we thoroughly consider the effect of sybils’ attack on badness propagation, which has been investigated in previous work rarely; in the second subroutine, we design two methods to convert the badness into peer's RA: the disproportional way and the logistic way. The second phase is the integration of RA into adjacent matrix used to represent P2P trust graph, to reflect peer's RA in trust propagation. The simulation results show that the disproportional and logistic DoubleFace can be robust against sybils’ and front peers’ attacks to reputation ranking and can achieve significant performance improvement in comparison with Eigentrust-like algorithms, in hostile P2P environments. Moreover, we also discuss the effect of hostile and hospitable P2P environments on the performance of DoubleFace. Specifically, disproportional DoubleFace can perform slightly better than logistic DoubleFace in hostile P2P environments, but, in hospitable P2P environments, disproportional DoubleFace performs anti-intuitively poorly, and logistic DoubleFace achieves the same performance as traditional Eigentrust-like algorithms.

In overlay networks, the interplay between network structure and dynamics remains largely unexplored. In this paper, we study dynamic coevolution between individual rational strategies (cooperative or defect) and the overlay network structure, that is, the interaction between peer's local rational behaviors and the emergence of the whole network structure. We propose an evolutionary game theory (EGT)-based overlay topology evolution scheme to drive a given overlay into the small-world structure (high global network efficiency and average clustering coefficient). Our contributions are the following threefold: From the viewpoint of peers' local interactions, we explicitly consider the peer's rational behavior and introduce a link-formation game to characterize the social dilemma of forming links in an overlay network. Furthermore, in the evolutionary link-formation phase, we adopt a simple economic process: Each peer keeps one link to a cooperative neighbor in its neighborhood, which can slightly speed up the convergence of cooperation and increase network efficiency; from the viewpoint of the whole network structure, our simulation results show that the EGT-based scheme can drive an arbitrary overlay network into a fully cooperative and efficient small-world structure. Moreover, we compare our scheme with a search-based economic model of network formation and illustrate that our scheme can achieve the experimental and analytical results in the latter model. In addition, we also graphically illustrate the final overlay network structure; finally, based on the group selection model and evolutionary set theory, we theoretically obtain the approximate threshold of cost and draw the conclusion that the small value of the average degree and the large number of the total peers in an overlay network facilitate the evolution of cooperation.

In overlay networks, the interplay between network structure and dynamics remains largely unexplored. In this paper, we study dynamic coevolution between individual rational strategies (cooperative or defect) and the overlay network structure, that is, the interaction between peer's local rational behaviors and the emergence of the whole network structure. We propose an evolutionary game theory (EGT)-based overlay topology evolution scheme to drive a given overlay into the small-world structure (high global network efficiency and average clustering coefficient). Our contributions are the following threefold: From the viewpoint of peers' local interactions, we explicitly consider the peer's rational behavior and introduce a link-formation game to characterize the social dilemma of forming links in an overlay network. Furthermore, in the evolutionary link-formation phase, we adopt a simple economic process: Each peer keeps one link to a cooperative neighbor in its neighborhood, which can slightly speed up the convergence of cooperation and increase network efficiency; from the viewpoint of the whole network structure, our simulation results show that the EGT-based scheme can drive an arbitrary overlay network into a fully cooperative and efficient small-world structure. Moreover, we compare our scheme with a search-based economic model of network formation and illustrate that our scheme can achieve the experimental and analytical results in the latter model. In addition, we also graphically illustrate the final overlay network structure; finally, based on the group selection model and evolutionary set theory, we theoretically obtain the approximate threshold of cost and draw the conclusion that the small value of the average degree and the large number of the total peers in an overlay network facilitate the evolution of cooperation.

Current information networks acting as the fundamental infrastructure of our society, possess the economic-social characteristics, so, in formulating new definitions and computational models for the networked environment, it is imperative to take economic and incentive considerations into account. The paper’s contribution is twofold: first, to characterize the economic implication of some proposed centrality in weighted network, we design the VCG (Vickrey-Clarke-Groves) overpayment based centrality in bi-connected networks, and compare it with existing global efficiency based centrality. Our experiments on weighted scale-free networks and small-world networks show the high correlation between global efficiency based centrality and VCG-based centrality (the Pearson correlation coefficients exceed 0.95); Then, inspired by the definition of global efficiency based centrality, we propose local efficiency based centrality, which, unlike global efficiency based centrality, can be calculated locally, and illustrates the effect of the proposed metric on the attack vulnerability of those weighted networks through comparing with strength-based attack.

Due to the fact that future wireless network will support multiclass services, so it is imperative to differentially treat various kinds of traffic types. An adaptive wireless resource management mechanism based on priority is proposed in this paper. Firstly, the mechanism estimates the reserved resource that is needed to support the necessary QoS in terms of handoff call dropping probability, then sets the thresholds according to the priority of various traffic types. The differentiated admission control scheme satisfies QoS requirement of various traffic types and admits high-priority traffic preferentially under high traffic load. The scheme proposed in this paper is simple and adaptive. Finally the paper presents the Markov model of this differentiated admission control, uses the numerical method to analyze the performance of this scheme, and compares the numerical results with simulation data. The results match relatively well.