Wireless Algorithms, Systems, and Applications

This paper addresses the problem of downlink channel estimation in millimeter wave (mmWave) massive multiple input multiple output (MIMO)-orthogonal frequency division multiplexing (OFDM) systems, where wideband frequency selective fading channels are considered. By exploiting the sparse scattering nature of mmWave channel, we consider channel estimation as three dimensional (3D) (including angles of departure/arrival and the time delay) line spectrum estimation. To achieve super-resolution channel estimation, we propose a decomposable 3D atomic norm minimization estimation method. This method decomposes the 3D estimation problem into two separate dimensions to reduce the computational complexity, where time delays are estimated only in the OFDM system. Simulation results show that the proposed method can achieve comparable mean square errors as the conventional vectorized ANM at much lower computational complexity.

In many IoT applications, sensor nodes are distributed over a region of interests and collect data at a specified time interval. With the development of hardware, the monitoring tasks become diversity. The specified acquisition strategy can not adaptively adjust the sampling interval. Due to the measurement error and the uncertainty of the environment, equi-frequency sampling technique may result in misunderstandings to the physical world. Based on Taylor expansion and time series analysis, this paper presents a sensed data model. The model can be considered as a unified approach, where linear regression or spline interpolation is a special case of our model. A mathematical method for parameter estimation is proposed, which can minimize the measurement error. And we prove the estimation is unbias. An adaptive data acquisition algorithm is proposed. Performance evaluation on the real data set verifies that the proposed algorithms have high performance in terms of accuracy and effectiveness.

The development of micro-energy harvesting technology provides a new energy solution for wireless sensor nodes (WSNs). Due to the intermittent power supplied by single environmental energy source, this paper proposes a hybrid energy harvesting architecture that harvest magnetic field (50–60 Hz) and solar energy simultaneously, which aims to provide a sustainable power supply for WSNs. Firstly, the design of free-standing “I-shaped” magnetic field transducer is introduced, which can harvest 0.17–0.46 mW underneath 700 A power transmission line. A further design of a rectifier and matching circuit is conducted and the maximum power point (MPP) of the hybrid energy harvesting circuit is about 60% of the open circuit voltage and the conversion efficiency reaches 61.68%. The experimental results show that the hybrid solar and “I-shaped” transducer can accomplish “cold start” operation of the power management unit (PMU) under magnetic flux density of 4.5 μT and light intensity of 200 lx, which will also provide a promising supply of energy for WSNs.

Along with the popularity of mobile devices, people share a growing amount of personal data to a variety of mobile applications for personalized services. In most cases, users can learn their data usage from the privacy policy along with the application. However, current privacy policies are always too long and obscure to provide readability and comprehensibility to users. To address this issue, we propose an automated privacy policy extraction system considering users’ personal privacy concerns under different contexts. The system is implemented on Android smartphones and evaluated feedbacks from a group of users ( $$n=96$$ ) as a field study. Experiments are conducted on both our dataset, which is the first user privacy concern profile dataset to the best of our knowledge, and a public dataset containing 115 privacy policies with 23K data practices. We achieve 0.94 precision for privacy category classification and 0.81 accuracy for policy segment extraction, which attests to the significance of our work as a direction towards meeting the transparency requirement of the General Data Protection Regulation (GDPR).

The dissemination of the basic safety messages (BSMs) is critical to ensure the safety performance of the vehicular networks. The wireless access in vehicular environment (WAVE) protocol restricts the BSM dissemination on the common control channel (CCH) which leads to a low dissemination efficiency. In this paper, we propose to allow the BSM dissemination on both the CCH and the service channels (SCHs). A cooperative BSM dissemination scheme, called as TLMV, is proposed for cooperator selection. The vehicle with short access delay to the SCH, short transmission delay on the SCH, and large number of neighbors working on the SCH is selected as the cooperator to disseminate the BSM on the SCH. Taking into account the fact that a vehicle may have great dissemination efficiency on multiple SCHs, TLMV-M which allows a vehicle to help disseminating the BSM on multiple SCHs is proposed to further improve the dissemination efficiency. Simulation results indicate that our TLMV and TLMV-M significantly reduce the time cost for BSM dissemination and increase the number of vehicles received the BSM compared with the existing workload-balanced shortest processing time first (WSPT) scheme.

The efficiency and reliability are crucial indexes when a trust system is applied into Wireless Sensor Networks (WSNs). In this paper, an efficient and reliable Trusted Intrusion Detection System (TIDS) with double cluster heads for WSNs is proposed. Firstly, an intrusion detection scheme based on trust is discussed. The monitoring nodes are responsible for evaluating the credibility of Cluster Member (CM) instead of depending on the feedback between CMs, which is suitable for decreasing the energy consumption of WSNs and reducing the influence of malicious nodes. Secondly, a new trust evaluation method is defined in TIDS and it takes the data forwarding and communication tasks into consideration which may enhance the reliability of Cluster Head (CH). The theoretical and simulation results show that our solution can effectively reduce the system overhead and improve the robustness of WSNs.

Sensitive data sharing through cloud storage environments has brought varies and flexible secure demands. Attribute-based signcryption (ABSC) is suitable for cloud storage because it provides combined data confidentiality and authentication, and fine-grained data access control. While, the existed ABSC schemes hardly support efficient attribute revocation. In addition, the heavy computational overhead of ABSC limits the applying resource-constrained device in cloud storage environments. In this paper, to tackle the above problems, we propose an efficiently revocable attribute-based signcryption scheme with decryption outsourcing. The proposed scheme achieves the efficient attribute revocation through delegating the cloud server to update ciphertext without decrypting it. During the decryption phase, it outsource massive decryption operations to the proxy server so that computation cost on user’s devices is small and constant. The security analysis proves the correctness, confidentiality, collusion resistance, unforgeability and forward secrecy of our scheme. Furthermore, performance analysis shows that our scheme is efficient in terms of the ciphertext, key size and computation cost while realizing desired functions.

This paper proposes a novel cognitive cooperative transmission scheme by exploiting massive multiple-input multiple-output (MMIMO) and non-orthogonal multiple access (NOMA) radio technologies, which enables a macrocell network and multiple cognitive small cells to cooperate in dynamic spectrum sharing. The macrocell network is assumed to own the spectrum band and be the primary network (PN), and the small cells act as the secondary networks (SNs). The secondary access points (SAPs) of the small cells can cooperatively relay the traffic for the primary users (PUs) in the macrocell network, while concurrently accessing the PUs’ spectrum to transmit their own data opportunistically through MMIMO and NOMA. Such cooperation creates a “win-win” situation: the throughput of PUs will be significantly increased with the help of SAP relays, and the SAPs are able to use the PU’s spectrum to serve their secondary users (SUs). The interplay of these advanced radio techniques is analyzed in a systematic manner, and a framework is proposed for the joint optimization of cooperative relay selection, NOMA and MMIMO transmit power allocation, and transmission scheduling. Further, to model network-wide cooperation and competition, a two-sided matching algorithm is designed to find the stable partnership between multiple SAPs and PUs. The evaluation results demonstrate that the proposed scheme achieves significant performance gains for both primary and secondary users, compared to the baselines.

Controlling a weighted linear dynamic network is important to various real world applications such as influencing political elections through a social network. Extant works mainly focus on minimizing the number of controllers that control the nodes in a network, but ignore the cost of controlling an individual node. Apparently, controlling a journalist or a mayor in a city has largely different costs, and we show that the aggregated control cost in extant works is often prohibitive. In this paper, we formulate the minimum control cost (MCC) problem in a weighted linear dynamic network, which is to find the set of controlled nodes with minimum sum of control costs. We show that the MCC problem is NP-hard by reducing the set cover problem to it. We also derive the lower/upper bounds and propose two approximation algorithms. Extensive evaluation results also show that the proposed algorithms have good performance compared to the derived lower bound of the problem.

In the era of context-aware services, users are enjoying remarkable services based on data collected from a multitude of users. However, in order to benefit from these services, users are enduring the risk of leaking private information. Game theory is a powerful method that is utilized to balance such tradeoff problems. The drawback is that most schemes consider the tradeoff problem from the aspect of the users, while the platform is the party that dominates the interaction in reality. There is also an oversight to formulate the interaction occurring between multiple users, as well as the mutual influence between any two parties involved, including the user, platform and adversary. In this paper, we propose a platform-centric two-layer three-party game model to protect the users’ privacy and provide quality of service. One layer focuses on the interactions among the multiple asymmetric users and the second layer considers the influence between any two of the three parties (user, platform, and adversary). We prove that the Nash Equilibrium exists in the proposed game and find the optimal strategy for the platform to provide quality service, while protecting private data, along with interactions with the adversary. Using real datasets, we present simulations to validate our theoretical analysis.

The APPs running on smart devices have greatly enriched people’s lives. However, they are collecting personally identifiable information (PII) secretly. The unrestricted collection, processing and unsafe transmission of PII will result in the disclosure of privacy, which cause losses to users. With the advent of laws and regulations about data privacy such as GDPR, the major APP vendors have become more and more cautious about collecting PII. However, the researches on detecting privacy leakage under GDPR framework still receive less attention. In this paper, we analyze the clauses of GDPR about privacy processing and propose a method for PII leakage detection based on Association Mining. This method assists us to find many hidden privacy leakages in traffic data. Moreover, we design and implement an automated system to detect whether the traffic data sent by the APPs reveals users’ PII. We have tested 509 APPs of different categories in the Google Play Store. The result shows that 76.23% of the APPs would collect and transmit PII insecurely and 34.06% of them would send PII to third parties.

Cross-domain roaming in Wi-Fi networks is ubiquitous and the frequency of global roaming of users has increased dramatically in recent years. To ensure network security, it is important to authenticate users belonging to different domains. Existing solutions like eduroam leverage a centralized and hierarchical architecture to authenticate users, which leads to serious performance and security issues in practice. In this paper, we propose Trustroam, a novel cross-domain authentication scheme in Wi-Fi networks based on blockchain. Different from traditional hierarchical solutions, Trustroam authenticates users and servers in a distributed and anonymous manner, avoiding several serious problems such as single point of failure and privacy leakage. Through the distributed consensus mechanism and mutual authentication, our scheme is highly fault tolerant to handle compromised server attacks. We implemented the Trustroam prototype in a real testbed. Experimental and evaluation results show that our scheme is superior to existing hierarchical solutions in terms of scalability, security and privacy preserving. Besides, Trustroam is an effective solution that can be conveniently and incrementally deployed in practical environments.

The penetration of Wireless Sensor Networks (WSNs) in various applications poses a high demand on reliable data storage, especially considering sensor networks are usually deployed in harsh environment. In this paper, we introduce Heterogeneous Wireless Sensor Networks where robust storage nodes are deployed in sensor networks and data redundancy is utilized through coding techniques, in order to improve the reliability of data storage. Taking into account the cost of both data delivery and storage, we propose an algorithm to jointly optimize data routing and storage node deployment. This problem is a binary non-linear combinatorial optimization, and it is highly non-trivial to design efficient algorithms due to its NP-hardness. By levering the Markov approximation framework, we elaborately deign a Continuous Time Markov Chain (CTMC) based scheduling algorithm to drive the storage node deployment and the corresponding routing strategy. Extensive simulations are performed to verify the efficacy of our algorithm.

Real-time identifying and tracking monitored objects is an important application in a public safety scenario. Both Radio Frequency Identification (RFID) and computer vision are potential solutions to monitor objects while faced with respective limitations. In this paper, we combine RFID and computer vision to propose a hybrid indoor tracking system, which can efficiently identify and track the monitored object in the scene with people gathering and occlusion. In order to get a high precision and robustness trajectory, we leverage Dempster-Shafer (DS) evidence theory to effectively fuse RFID and computer vision based on the prior probability error distribution. Furthermore, to overcome the drift problem under long-occlusion, we exploit the feedback from the high-confidence tracking results and the RFID signals to correct the false visual tracking. We implement a real-setting tracking prototype system to testify the performance of our proposed scheme with the off-the-shelf IP network camera, as well as the RFID devices. Experimental results show that our solution can achieve 98% identification accuracy and centimeter-level tracking precision, even in long-term occlusion scenarios, which can manipulate various practical object-monitoring scenarios in the public security applications.

Video-based person re-identification (ReID) is a problem of person retrieval that aims to match the same person in two different videos, which has gradually entered the arena of public security. The system generally involve three important parts: feature extraction, feature aggregation and loss function. Pedestrian feature extraction and aggregation are critical steps in this field. Most of the previous studies concentrate on designing various feature extractors. However, these extractors cannot effectively extract spatiotemporal information. In this paper, several spatiotemporal convolution blocks were proposed to optimize the feature extraction model of person Re-identification. Firstly, 2D convolution and 3D convolution are simultaneously used on video volume to extract spatiotemporal feature. Secondly, non-local block is embedded into ResNet3D-50 to capture long-range dependencies. As a result, the proposed model could learn the inner link of pedestrian action in a video. Experimental results on MARS dataset show that our model has achieved significant progress compared to state-of-the-art methods.

Side channels are actively exploited by attackers to infer users’ privacy from publicly-available information on Android devices, where attackers probe the states of system components (e.g., CPU and memory), APIs, and device sensors (e.g., gyroscope and microphone). These information can be accessed by applications without any additional permission. As a result, traditional permission-based solutions cannot efficiently prevent/detect these probing attacks. In this paper, we systematically analyze the Android side-channel probing attacks, and observe that the high frequency sensitive data collecting operations from a malicious app caused continuous changes of its process states. Based on this observation, we propose SideGuard, a process-state-based approach to detect side-channel probing attacks. It monitors the process states of the applications and creates the corresponding behavior models described by feature vectors. Based on the application behavior models, we train and obtain classifiers to detect malicious app behaviors by using learning-based classification techniques. We prototyped and evaluated our approach. The experiment results demonstrate the effectiveness of our approach.

Modern applications in mobile computing become increasingly complex and computation intensive. Task offloading from mobile devices to the cloud is more and more frequent. Edge Computing, deploying relatively small-scale edge servers close to users, is a promising cloud computing paradigm to reduce the network communication delay. Due to the limited capability, each edge server can be configured with only a small amount of functions to run corresponding tasks. Moreover, a mobile application might consist of multiple dependent tasks, which can be modeled and scheduled as Directed Acyclic Graphs (DAGs). When an application request arrives online, typically with a deadline specified, we need to configure the edge servers and assign the dependent tasks for processing. In this work, we jointly tackle on-demand function configuration on edge servers and DAG scheduling to meet as many request deadlines as possible. Based on list scheduling methodologies, we propose a novel online algorithm, named OnDoc, which is efficient and easy to deploy in practice. Extensive simulations on the data trace from Alibaba (including more than 3 million application requests) demonstrate that OnDoc outperforms state-of-the-art baselines consistently on various experiment settings.

Using magnetic beamforming, the power transmission efficiency can be highly increased for realizing the practical near-field magnetic resonant coupling (MRC) wireless power transfer (WPT). The usage of relay coils is also of great benefit for power transmission in the WPT system containing multiple transmitters (TXs) each with a coil. In this paper, we study the influence of relay coils and the magnetic beamforming for a MRC-WPT system with multiple TXs, relay coils and a single receiver (Rx), called multi-relay WPT (MWPT) system. The optimization problem is formulated to design the currents flowing through TXs in different places so as to maximize the power delivered to the RX load, subject to a given constraint of summational power consumed by all resistances in the MWPT system. In general, the problem is a non-convex quadratically constrained quadratic programming (QCQP) problem, which can be solved with relaxation of the Lagrange multiplier. Numerical results show that by comparing with equal-current and magnetic beamforming mode, relay coils significantly enhances the power transmission efficiency over the longer distances, and magnetic beamforming with passive relay coils significantly improve the transmission power, efficiency and distance.

With the rapid development of e-commerce, various types of recommendation systems have emerged in an endless stream. Collaborative filtering based recommendation methods are either based on user similarity or item similarity. Neural network as another choice of recommendation method is also based on item similarity. In this paper, we propose a new model named Self Attention based Collaborative Neural Network (SATCoNN) to combine both user similarity and item similarity. SATCoNN is an extension of Recurrent Neural Network (RNN). SATCoNN model uses self-attention mechanism to compute the weight of products in multi aspects from user purchase history which form a user purchase history vector. Borrowing the idea of image style transfer, we model the users’ shopping style by gram matrix. We exploit the max-pooling technique to extract users style as a style vector in gram matrix. The experimental results show that our model has better performance by comparison with other recommendation algorithms.

With data growing exponentially, more and more people prefer to share data with others by storing the data in edge servers. However, edge server cannot be deemed completely trustable as the sensitive data my be disclosed. Therefore, in this paper, we propose an efficient and secure data sharing scheme for edge storage by employing Ciphertext-Policy Attribute-Based Encryption (CP-ABE) which can be utilized to conduct fine-grained control. This scheme can not only support data recovery when some edge servers break down by employing Secret Sharing Scheme, but also can support semi-trusted third party authority via employing re-encryption method. That is, the third party authority can not either reveal the private data stored in edge servers. Finally, we analyze security of our scheme to demonstrate that this scheme is resistant to eavesdropping attack and colluding attack. Additionally, relevant experiments results are shown that the scheme is feasibility and efficiency.

This paper investigates the truthfulness establishment problem between two nodes (vehicles) in a vehicular network. We focus more on the case when no interaction has been conducted and we use the Point of Interests (POIs) visited by the two nodes (vehicles) to establish the initial truthfulness. It turns out that this is a general version of a well-studied problem in computational genomics called CMSR (Complementary Maximal Strip Recovery) in which the letters (similar to POIs) cannot be duplicated, while in our problem POIs could certainly be duplicated. We show that one version (when noisy POIs are deleted all the remaining POIs must be involved in some adjacency), is NP-hard; while the other version (with the adjacency involvement constraint is dropped), is as hard as Set Cover. We then design a practical solution based on local search for the first problem. Simulations with various synthetic data show that the algorithm is very effective.

Continuous publication of statistics over user-generated streams can provide timely data monitoring and analysis for various applications. Nonetheless, such published statistics may reveal the details of individuals’ sensitive status or activities. To guarantee the privacy for event occurrences in data streams, based on the known privacy standard of $$\varepsilon $$ -differential privacy, w-event privacy has been proposed to hide multiple events occurring at continuous time instances. Nonetheless, the too strict requirement of w-event privacy makes it hard to achieve effective privacy protection with high data utility in many real-world scenarios. To this end, in this paper we propose a novel notion of average w-event privacy and the first Lyapunov optimization-based privacy-preserving scheme on infinite streams, aiming to obtain higher data utility while satisfying a relatively stable privacy guarantee for whole streams. In particular, we first formulate both our proposed privacy definition and the utility loss function of statistics publishing in a stream setting. We then design a Lyapunov optimization-based scheme with a detailed algorithm to maximize the publishing data utility under the requirement of our privacy notion. Finally, we conduct extensive experiments on both synthetic and real-world datasets to confirm the effectiveness of our scheme.

Due to their high performance as well as their low design complexity, polar codes are being considered as a candidate for next generation of mobile and wireless communications. Originally, polar codes were exactly designed over binary erasure channels (BECs) only. Later, polar codes over additive white Gaussian noise (AWGN) channel and fading channels were discussed. Some researches have investigated the performance of polar codes over multipath fading environment characterized by Rayleigh model which, however, fails to accurately predict wireless environments of high frequencies and long distance transmissions. The performance analysis of polar codes over a practical set is rarely addressed. To this end, this paper is devoted to investigate the performance of polar codes over a fading environment characterized by Nakagami-m fading model which demonstrates closer estimates to the measurements of real wireless channels. Then, we investigate the performance of polar codes over real channel measurements collected in an indoor as well as V2V communication environments. The remarkable error rate performance of polar code shows that it will be beneficial to future wireless systems where high data rates and low BERs are necessary.

With the popularity of smart phones and the maturity of civilian global positioning system (GPS) technology, transportation network company (TNC) services have become a prominent commute mode in many major cities, which can effectively pair the passengers with the TNC vehicles/drivers through mobile applications. However, given the growing number of TNC vehicles, how to efficiently dispatch TNC vehicles poses crucial challenges. In this paper, we propose a novel method for TNC vehicle dispatching in different areas of the city based on deep reinforcement learning (DRL) method with joint consideration of the TNC company, individual TNC vehicle, and customer/passenger. The proposed model optimizes the distribution of vehicles geographically to meet the customers’ demands, while improving the drivers’ profit. In particular, we consider the high dimensional state and action space in the urban city traffic dynamic environment, and develop a deep deterministic policy gradient, an actor-critic based DRL algorithm for dispatching vacant TNC vehicles. We leverage Didi Chuxing’s open data set to evaluate the performance of the proposed approach, and the simulation results show that the proposed approach improves the average income of the driver while satisfying the supply and demand relationship between TNC vehicles and customers/passengers.

With the proliferation of sensor-embedded mobile devices, mobile crowdsensing has become a paradigm of significant interest. Incentivizing sensory-data providers to keep sustainability in a mobile crowdsensing system is a critical issue nowadays, and auction-based mechanisms have been proposed to motivate providers via monetary rewards. In our work, this sustainability problem is formulated as an optimization problem maximizing providers’ proportionally fair utilities with respect to their multi-dimensional fairness factors, and a fairness-aware auction mechanism is designed accordingly. To the best of our knowledge, this is the first work that considers multi-dimensional fairness of providers as the objective in selecting providers for the mobile crowdsensing system. In addition, we present rigorous theoretical analysis proving that our mechanism meets budget feasibility, individual rationality and truthfulness. Finally, simulations are performed to demonstrate the performance of our proposed mechanism.

Considering performance and safety of cyber-physical system (CPS), data transmission between devices arranged within the same working area and outside network must be achieved by gateways. Therefore, to guarantee the reliability of the system, one working area is often covered by multiple gateways, and each gateway has limited bandwidth. For its own benefit, each device intends to occupy as much bandwidth as possible. That will occur imbalance and degrade performance of the system. In this paper, we propose an evolutionary game based gateway selection algorithm which can guarantee the fairness among devices in CPS. The load balancing of gateways can also be achieved by using this algorithm. By using evolutionary game theory, we analyze the behaviors of devices when they obtained less bandwidth than the average. The bandwidth allocation model of gateways has been proposed, and we formulate the gateway switching procedure of devices as an evolutionary game. We propose the replicator dynamics of this evolutionary game and analyze the existence and stabilization of the evolutionary equilibrium. Simulation results show that the proposed algorithm converges fast and minimize the maximal difference between any two devices of the same kind.

A smart grid is a complex system using power transmission and distribution networks to connect electric power generators to consumers across a large geographical area. Due to their heavy dependencies on information and communication technologies, smart grid applications, such as state estimation, are vulnerable to various cyber-attacks. False data injection attacks (FDIA), considered as the most severe threats for state estimation, can bypass conventional bad data detection mechanisms and render a significant threat to smart grids. In this paper, we propose a novel FDIA detection mechanism based on a wide and recurrent neural networks (RNN) model to address the above concerns. Simulations over IEEE 39-bus system indicate that the proposed mechanism can achieve a satisfactory FDIA detection accuracy.

Client-independent Internet Protocol address (IP) geolocation is a critical problem in the Internet World, of which the accuracy is based on highly reliable landmarks. However, most existing methods focus heavily on improving the location estimating method rather than improving the quality and quantity of landmarks. Without sufficient landmarks of high quality, they face difficulties when attempting to further improve accuracy. Even though some existing mining based methods dig massive landmarks from online web resources, most landmarks are of low quality because they do not make full use of these open resources. In this paper, we propose ONE-Geo, a methodology to mine highly reliable landmarks as much as possible by extracting the owner name of web servers. For a given target IP, ONE-Geo extracts the real owner name from web page information and registration records. Utilizing this clue, ONE-Geo determines the correct location by searching address information on an organization knowledge graph and conduct inference. Experimental results show that ONE-Geo achieves a median error distance of 463 m on 165 web servers and a median error distance of 7.7 km on 721 nodes that do not host a website. For web servers, ONE-Geo outperforms existing methods and several commercial tools. To be specific, 66.1% nodes are geolocated by ONE-Geo with an error less than 1 km, which is two times as many as Street-level Geolocation(SLG), which is one of the best existing methods on IP geolocating.

In Bitcoin, the knowledge of private key equals to the ownership of bitcoin, which occurs two problems: the first problem is that the private key must be kept properly, and the second one is that once the private key is given, it can’t be taken back, hence the bitcoin system can only implement the transfer function. In this paper, we first propose a new digital signature algorithm and use it to design an online wallet, which can help the user derive the signature without obtaining the user’s private key. Secondly, using our proposed online wallet, we extend the application of private key so that the cryptocurrency system can implement the authorization function. In more detail, we define a new primitive that we call decentralized hierarchical authorized payment scheme (DHAP scheme). We next propose a concrete instantiation and prove its correctness. Finally, we analyze the security and usability of our scheme. For security, we prove our scheme to be secure under the random oracle model. For usability, we examine its performance and compare it with bitcoin’s performance.

High Altitude Platforms Station (HAPS) is considered to be an effective solution to expand the communication coverage of rural area in the fifth generation (5G) network. However, HAPS is usually in an unstable state because of space airflow. Thus, the inaccurate beamforming performed by the gateway (GW) will result in unnecessary capacity loss of HAPS communication system. To address this issue, a long short-term memory (LSTM)-based location predictive model is proposed to predict next moment location of HAPS by training the current two-dimensional (2D) angle data. Specifically, a novel preprocessing system is introduced to ensure the effectiveness of our model. Moreover, the LSTM-based model with highest predictive accuracy can be saved during the training to realize the real-time prediction. Experimental results reveal that the proposed LSTM-based model is of higher prediction accuracy compared with other two predictive models. Therefore, a more precise beamforming performed by GW can reduce the unnecessary capacity loss and improve the reliability of 5G HAPS communication system.

Communication interference and the energy limitation of nodes seriously hamper the fundamental performance of wireless sensor networks (WSN) such as throughput and network lifetime. In this paper, we focus on the Successive Interference Cancellation (SIC) aiming to realize multi-node concurrency communication and propose a heuristic power control algorithm. To prolong the network lifetime, we consider the scenario of mobile wireless charging equipment (WCE) periodically charging each node’s battery wirelessly. Time-slice scheduling scheme and energy consumption optimization protocol are adopted to design an efficient cross-layer charging strategy. Then we use a near-optimal method to transform the original problem into a linear problem which yields identical optimal value. Simulation results demonstrate that adopting SIC and WCE can greatly improve channel utilization ratio and increase network throughput by 200% to 500% while ensuring the network lifetime.

An important part of environmental monitoring is the collection of meteorological data. In this paper, we develop an integrated data acquisition and transmission platform utilizing unmanned aerial vehicles (UAVs) with an intelligent path planning algorithm to achieve efficient and accurate meteorological data collection in real-time. We adopt the improved traveling salesman problem model to represent the path planning problem. Based on the model, we propose an improved simulated annealing genetic algorithm (ISAGA) to solve the path planning problem. Our proposed ISAGA is able to overcome the deficiencies of the traditional genetic algorithm and simulated annealing algorithm. In addition, we design and implement a mobile application integrated with the path planning algorithm to control UAVs and conduct data exchange to the cloud. Our evaluation results demonstrate that data can be collected and transmitted more efficiently via selecting better paths.

Diagnosing multiple lesions on images is facing with challenges of incomplete and incorrect disease detection. In this paper, we propose a deep model called CXNet-m2 for the detection of multiple lesions on chest X-ray images. In our model, there is a convolutional neural network (CNN) for encoding the images, a recurrent neural network (RNN) for generating the next word (the name of lesion) and an attention mechanism to align the visual contexts with the prediction of words. There are two main contributions of CXNet-m2 to improve the work efficiency and increase the diagnosis accuracy. (1) Inspired by image captioning, CXNet-m2 adapts the classification system to a language model, where Bi-LSTM is used to learn the clinical relationship between lesions. (2) Inspired by attention mechanism, the prediction of possible lesions is guided by visual contexts, where the visual contexts are selected by the previously generated words and chosen visual regions.The experimental results on Chestx-ray14 show that CXNet-m2 achieves better AUC and the different versions of CXNet-m2 illustrate the importance of pre-training and clinical contexts.

We study a classical opportunistic wireless link scheduling problem in cognitive radio networks with Signal to Interference plus Noise Ratio (SINR) constraints. Consider a collection of communication links, assume that each link has a channel state. The state transitions follow a transition rule. The exact state information of each link is not available due to the uncertainty of primary users’ activities. The expected channel state is predicted probabilistically by investigating its history and feedbacks when the channels are used. The objective is to pick communication links sequentially over a long time horizon to maximize the average reward. To the best of our knowledge, no prior work can satisfyingly provide solutions for the opportunistic wireless link scheduling problem when considering SINR constraints. In this work, we adopt the robust paradigm of restless multi-armed bandit for the problem and design an efficient algorithm. We analyze the performance via Lyapunov potential function and demonstrate that the proposed algorithm can achieve an approximation bound.

Wireless sensor network (WSN) systems often need to support real time periodic queries of physical environments. In this work, we focus on periodic queries with sufficiently long time horizon in duty-cycled sensor networks. For each periodic query issued by a control center in a WSN, after the source sensors produced the sensory data, the data are to be sent to the sink via multi-hop data aggregation timely in a periodic fashion. To this end, we propose efficient and effective data aggregation algorithms subject to quality of service constraints such as deadline requirements and interference constraints. We decompose these into three sequential operations: (1) aggregation tree construction (2) node and link-level scheduling and (3) packet scheduling. Inspired by the scheduling algorithms, we identify both sufficient conditions and necessary conditions for scheduling multiple queries. The schedulability analysis under various interference models demonstrate that the proposed algorithms achieve an approximate proportion of the maximum possible load.

Edge computing enables the data to be processed in the edge of networks in order to decrease the latency of crowdsensing services. However, due to the distributed environment and vulnerability of edges, it is difficult for different edges to reach consistency to provide the same service and protect the data from tampering at the same time. To solve these problems, the Blockchain, a credible and natural decentralized technique, is considered as a suitable tool. In this paper, we proposed a Blockchain-based edge crowdsensing service system in which the edge runs a changeable auction algorithm for every task that the users request to find a winner who can provide corresponding sensing data. Specifically, based on PBFT algorithm, we proposed a consensus algorithm named Leader Stable Practical Byzantine Fault Tolerance (LS-PBFT). This algorithm enables all edges to collaboratively maintain an updated, consistent and credible ledger in Blockchain. Furthermore, the data generated in this process are constructed as a multi-transaction, which can be packaged into a block and stored in the block. Simulation results reveal that the proposed system is not only efficient in generating and storing blocks but also feasible in resisting attacks of malicious users and edges. Our experiments also show that LS-PBFT takes less than 50% of the time cost by PBFT to reach consensus.

Online social networks provide platforms for people to interact with each other and share moments of their daily life. The online social network data are valuable for both academic and business studies, and are usually processed by anonymization methods before being published to third parties. However, several existing de-anonymization techniques can re-identify the users in anonymized networks. In light of this, we explore the impact of user attributes in social network de-anonymization in this paper. More specifically, we first quantify the significance of attributes in a social network, based on which we propose an attribute-based similarity measure; then we design an algorithm by exploiting attribute-based similarity to de-anonymize social network data; finally we employ a real-world dataset collected from Sina Weibo to conduct experiments, which demonstrate that our design can significantly improve the de-anonymization accuracy compared with a well-known baseline algorithm.

The fault diagnosis in a modern communication system is traditionally supposed to be difficult, or even impractical for a purely data-driven machine learning approach, for it is a humanmade system of intensive knowledge. A few labeled raw packet streams extracted from fault archive can hardly be sufficient to deduce the intricate logic of underlying protocols. In this paper, we supplement these limited samples with two inexhaustible data sources: the unlabeled records probed from a system in service, and the labeled data simulated in an emulation environment. To transfer their inherent knowledge to the target domain, we construct a directed information flow graph, whose nodes are neural network components consisting of two generators, three discriminators and one classifier, and whose every forward path represents a pair of adversarial optimization goals, in accord with the semi-supervised and transfer learning demands. The multi-headed network can be trained in an alternative approach, at each iteration of which we select one target to update the weights along the path upstream, and refresh the residual layer-wisely to all outputs downstream. The actual results show that it can achieve comparable accuracy on classifying Transmission Control Protocol (TCP) streams without deliberate expert features. The solution has relieved operation engineers from massive works of understanding and maintaining rules, and provided a quick solution independent of specific protocols.

Wireless sensor networks (WSNs) are constrained by limited battery energy and channel utilization. Thus, ephemeral network lifetime and low channel utilization are widely regarded as the performance bottlenecks. In this paper, we investigate the operation of multi-hop recharging sensor wireless networks (WRSNs) with successive interference cancellation (SIC) technique. In WRSNs, the power of the sensor nodes are not constants, but the receiving power of nodes need to be sorted. To solve the problem, we first establish a minimum energy routing, unify nodes transmission rate for determining the power of transmitting and time Schedule scheme. Then, we analyze an optimization problem with objective of maximizing the mobile charger’s (MC) vacation time over the cycle time. Subsequently, we develop a near-optimal solution and verify its feasible performance. Simulation results show that SIC can achieve better throughput (about increasing 350%–550% compared with inference avoid) and no extra energy consumption in multi-hop WRSNs.

With the rapid popularity of smart mobile devices and the rapid development of location-based social networks (LBSNs), location-based recommendation has become an important method to help people find the attractive point-of-interest (POI). However, due to the sparsity of user-POI check-in data, the traditional recommendation model based on collaborative filtering cannot be well applied to the POI recommendation problem. In addition, location-based social networks are different from other recommendation scenarios, and users’ POI check-ins are closely related to social relations and geographical factors. Therefore, this paper proposes a neural networks POI recommendation model fused with social and geographical graph embedding representation(SG-NeuRec). Our model organically combines social and geographical graph embedding representations with user-POI interaction representation, and captures the latent interactions between users and POIs under the neural networks framework. Meanwhile, in order to improve the accuracy of POI recommendation, the relevance between users’ accessing time pattern and POI is modeled by the designed shallow network and unified under the same framework. Extensive experiments on two real location-based social networks datasets demonstrate the effectiveness of the proposed model.

In recent years, web crawlers have been widely used for collecting data from the Internet. Accurately recognizing web crawlers can help to better utilize friendly crawlers while stopping malicious ones. Existing web crawler recognition researches have difficulties in handling new crawlers, such as distributed crawlers, proxy based crawlers, and browser engine based crawlers. Moreover, it is non-trivial to achieve both high identification accuracy and high response time simultaneously. To tackle these issues, we propose a novel approach to web crawler recognition which combines real-time recognition methods based on heuristic rules and offline recognition methods based on machine learning. The aforementioned problems are well solved in this approach. The advantage of this approach is that both accuracy and efficiency are improved. We build a website and analyze its web access log using the proposed method. According to the results, the proposed approach achieves desirable performance in both accuracy and efficiency.

We design and analyze contextual combinatorial cascading Thompson sampling ( $$C^3$$ -TS). $$C^3$$ -TS is a Bayesian heuristic to balance the exploration-exploitation tradeoff in the cascading bandit model. And it incorporates the linear structure to share information among different items. These two important features allow us to prove an expected cumulative regret bound of the form $${\tilde{O}}(d\sqrt{KT})$$ , where d and K are the dimension of the feature space and the length of the chosen list respectively, and T is the number of time steps. This regret bound matches the regret bounds for the state-of-the-art UCB-based algorithms. More importantly, it is the first theoretical guarantee on a contextual Thompson sampling algorithm for cascading bandit problem. Empirical results demonstrate the advantage of $$C^3$$ -TS over existing UCB-based algorithms and non-contextual TS in terms of both the cumulative reward and time complexity.

In this paper, we investigate the problem of computing a consensus trajectory of a vehicle giving the history of Points of Interest (POIs) visited by the vehicle over certain period of time. The problem originates from building the social connection between two vehicles in a vehicular network. Formally, given a set of m trajectories (sequences $$S_i$$ ’s over a given alphabet $$\varSigma $$ , each with length at most O(n), with $$n=|\varSigma |$$ ), the problem is to compute a target (median) sequence T over $$\varSigma $$ such that the sum of similarity measure (i.e., number of adjacencies) between T and all $$S_i$$ ’s is maximized. For this version, we show that the problem is NP-hard and we present a simple factor-2 approximation. If T has to be a permutation, then we show that the problem is still NP-hard but the approximation factor can be improved to 1.5. We implement the greedy algorithm and a variation of it which is based on a more natural greedy search. Using simulated data over two months (e.g., $$m=60$$ ) and variants of $$|S_i|$$ and $$\varSigma $$ (e.g., $$30\le |S_i|\le 100$$ and $$30\le |\varSigma | \le 60$$ ), the empirical results are very promising and with the local adjustment algorithm the actual approximation factor is between 1.5 and 1.6 for all the cases.

Mobile edge computing (MEC) casts the computation-intensive and delay-sensitive applications of mobiles on the network edges. Task offloading incurs extra communication latency and energy cost, and extensive efforts have been focused on the offloading scheme. To achieve satisfactory quality of experience, many metrics of the system utility are defined. However, most existing works overlook the balancing between the throughput and fairness. This paper investigates the problem of seeking optimal offloading scheme and the objective of the optimization is to maximize the system utility for leveraging between throughput and fairness. Based on KKT condition, we analyze the expectation of time complexity for deriving the optimal scheme. We provide an increment based greedy approximation algorithm with $$1 + \frac{1}{{e - 1}}$$ ratio. Experimental results show that the proposed algorithm has better performance.

The internet of things (IoT) enables physical objects to sense the world and hence perform specific tasks, winning a great market success. Routing bears significant importance to IoT since punctual and reliable information delivery is indispensable for most IoT applications. The start-of-the-art work on IoT routing mainly focuses on designing algorithms to select efficient relays according to a given topology for improving routing performance. In this paper, we take a dramatically different viewpoint to study the routing in IoT. In detail, we figure out the best IoT network structure to attain the best message transmission. We have proved that IoT with small-world properties can achieve better routing performance when the probability of long-range connections obeys Lévy distribution. Furthermore, we deduce the upper and lower bound of the average number of message transmission steps respectively. Finally, we do extensive experiments to validate our analysis.

Most research on information propagation in social networks does not consider how to find information dissemination paths from the information source node to a set of influential nodes. In this paper, we introduce a multicast information propagation model which disseminates information from the information source node to a set of designated influential nodes in social networks, and formulate the problem with the objective to maximize the social influence on the information propagation paths. We then propose a Parallel Multicast information Propagation algorithm (PMP), which concurrently constructs a subgraph for each influential node, joins all the subgraphs into a merge graph, and finds the information propagation paths with the maximum social influence in the merge graph. The simulation results demonstrate that the proposed algorithm can achieve competitive performance in terms of the social influence on the information propagation paths.

Aiming at the physical layer information security risk in wireless communication system, a physical layer security transmission technology based on OAM (Orbital angular momentum) modulation is proposed in this paper. Through introducing the OAM state as an information bearing parameter, the modulated electromagnetic (EM) wave will have a helical transverse phase structure, which means that the phase front varies linearly with azimuthal angle. In such case, the modulated signal will present the direction-dependent characteristic; in addition, since the OAM state becomes very weak beyond a certain distance, utilizing this characteristic, the effective communication distance can be limited. Therefore, the communication range can be accurately limited by OAM modulation. Furthermore, an OAM pre-compensation algorithm is proposed to improve the QoS of legitimate users. The simulation results show that the proposed scheme can effectively prevent eavesdroppers from detecting OAM modulated signals.

Double-loop networks are widely used in computer networks for its simplicity, symmetry and scalability. In this paper, we focus on optimal routing of Bidirectional Double-loop Network (BDLN) using coordinates embedding and transforming. First, we get the lower bound both of diameter and average distance of BDLN by embedding BDLN into Cartesian coordinates. Then, we find nodes distribution regularity on the embedding graph of tight optimal BDLNs that achieve the lower bound both in diameter and average distance. On the basis of nodes distribution regularity in tight optimal BDLNs, we present on demand optimal message routing algorithms which do not require routing tables and are highly efficient requiring very little computation.

Online Social Network (OSN) has become an indispensable part of our daily life. Analyzing influence factors and propagation rules for rumor spreading in OSN is of great significance to the guidance and prediction of social public opinion. This paper proposes a dynamic information propagation model, Disseminate & Discriminate-Spread-Exposed-Ignorant-Recover (i.e., DDSEIR) based on user’s dissemination capacity and discriminant ability. First, this paper introduces a hierarchical mechanism based on degree centrality theory to roughly categorize Internet users according to user’s dissemination capacity. And subsequently, user’s discriminant ability can be evaluated by sender’s identity and information attributes. Finally, experimental results prove that the DDSEIR not only delays the time of rumor spreading but also reduces the scale of rumor propagation.

Vehicular Named Data Networking (V-NDN) is an Vehicular Ad-hoc Network (VANET) using Named Data Network (NDN). Interest packet forwarding and data packet caching strategy are two key issues in the field. This paper focus on the data distribution and caching strategy of V-NDN in urban road environment. First, we propose a RSU-assisted strategy for interest packet and data packet forwarding. Secondly, it is the first time to use the method of decision tree prediction to guide the data packet cache, and uses a cache replacement policy based on popularity and request cost to store data due to the memory limit of RSU. Finally, the simulation results show that the proposed strategy and method effectively improve quality of service (QoS) of network.

Internet of Things (IoT) devices are typically resource constrained micro-computers for domain-specific computations. Most of them use low-cost embedded system that lacked basic security monitoring and protection mechanisms. Consequently, IoT-specific malwares are made to target at these vulnerable devices for deep infection and utilization, such as Mirai and Brickerbot, which poses tremendous threats to the security of IoT. In this issue, we present a novel approach for detecting malware in IoT environments. The proposed method firstly extract one-channel gray-scale image sequence that converted from the disassembled malware binaries. Then we utilize a Two-Bits Convolutional Neural Network (TBN) for detecting IoT malware families, which can encode the network edge weights with two bits. Experimental results conducted on the collected dataset show that our approach can reduce the memory usage and improve computational efficiency significantly while achieving a considerable performance in terms of malware detection accuracy.

The emergence of the crowd sensing solves the problem that the traditional perception mode is hard to deploy on a large scale and at a high cost. However, users are exposed to the risk of privacy leakage when participating in crowd sensing. In order to solve this issue, this paper protects the user’s privacy through the dynamic group collaborative data submission mechanism and the method of adding noise perturbation, solves the privacy protection problem in the case of collusion attack. While implementing privacy protection and taking into consideration performance, this solution further reduces the cost of the system through batch verification. Safety analysis and simulation show the effectiveness and efficiency of the proposed method.

In the cloud storage, the data owner will lose the direct control over his outsourced data, and all the operations over the outsourced data may be executed by corresponding remote cloud server, such as cloud data deletion operation. However, the selfish cloud server might maliciously reserve the data copy for financial interests, and deliberately send a false deletion result to cheat the data owner. In this paper, we design an IBF-based publicly verifiable cloud data deletion scheme. The proposed scheme enables the cloud server to delete the data and return a proof. Then the data owner can check the deletion result by verifying the returned deletion proof. Besides, the proposed scheme can realize public verifiability by applying the primitive of invertible bloom filter. Finally, we can prove that our proposed protocol not only can reach the expected security properties but also can achieve the practicality and high-efficiency.

Recommender systems, which provide users with suggestions for selecting items that is of potential interest to them, are widely used to assist mobile users in reducing information overload and making better choice quickly in their daily life. Social recommender systems, which have the potential to mitigate the new user cold-start problem, utilize social relationships as an extra source of information. As recommendation results depend on users’ individual data, privacy breaches may occur. Although several differentially private social recommender systems have been proposed, their application scopes or protection strengths are limited. In this paper, we propose a differentially private social recommender system for mobile users named $$\mathcal D\!P\!S\!R$$ to block curious users from inferring the existence of someone else’s numeric rating or social relationship. Empirical evaluations on two real-world datasets are conducted, and the results show that $$\mathcal D\!P\!S\!R$$ can balance the utility of recommendations with the privacy of users’ data in both normal and cold-start test view.