Chainweb and some other parallel blockchain systems have recently been proposed, with the objectives of improving the throughput and enhancing the tamper-proof capability. While many security related studies have been conducted for traditional single-chain based blockchain systems, the security aspect of parallel chain systems is yet to be well studied and understood. Our paper presents a systematic study on selfish mining attacks in Chainweb based on mathematical modeling. Specifically, selfish mining is conducted by concentrating the computation power on a subset of parallel chains and operating a proper withholding strategy. We demonstrate how to establish a Markov chain based analytical model with innovative techniques to handle the very large state space. Our Markov chain model is also capable of handling different number of parallel chains. The mathematical analysis brings an insightful, in fact counterintuitive, finding that the attackers need less computation power to harvest additional rewards through withholding when Chainweb contains a larger number of chains; while the common understanding is that the more chains are used, the more tamper-proof the system is. The accuracy of the Markov chain analysis is demonstrated via comparison to the simulation results.

Ensuring data freshness is important for Industrial Internet of Things (IIoT). In this article, we consider a typical IIoT application where multiple sensors monitor some time-varying physical processes and report measurements to a central base station through an unreliable wireless channel. In order to save energy, each sensor may switch to sleep mode for a while after successfully transmitting a packet. Since only awake sensors are able to transmit data, we propose a novel function of Age of Information with penalty (or AoI-penalty) to capture the eagerness of the active sensors to provide fresh information. We formulate a new AoI-penalty minimization problem for scheduling the sensors’ transmissions. We theoretically derive a necessary condition for the system’s AoI-penalty to converge to a finite value, and further obtain a lower bound of the AoI-penalty. Moreover, we develop a max-weight-based scheduling policy and theoretically prove that it is the optimal policy when the network is symmetric and the channel is error-free. Simulation results demonstrate that the proposed policy achieves AoI performance near to the lower bound and that with such sleep–wake sensors, the achieved AoI performance is close to that with nonsleeping sensors but at a much lower energy cost.

Data-driven machine learning approaches have been proposed to facilitate wireless network optimization by learning latent knowledge from historical optimization instances. However, existing works use simplistic network representations that cannot properly encode the topological difference. They are often limited to fixed topology, and the performance is degraded because the learning target does not get sufficient information since the topological information is not well captured.To address this, we leverage the graphical neural network techniques and propose a two-stage topology-aware deep learning (TADL) framework, which trains a graph embedding unit and a link usage prediction module jointly to discover links likely to be used in optimal scheduling. By properly encoding the network structure, it makes input data with varying topology possible, and also provides more informative clues for the learning target.Important techniques are developed to ensure learning efficiency. The performance is evaluated on canonical multi-hop flow problems with diverse network structures, sizes and realistic deployment scenarios. It achieves close-to-optimum solution quality with a significant reduction in computation time without retraining.

The next mobile generation is highly expected since it is supposed to increase the bit rate and reduce latency to allow multiple new services been offered. However, there is a big concern about energy efficiency, since, more base stations must be added, providing amplified coverage and higher spectral efficiency. There have been several algorithms designed to reduce the energy consumption of wireless networks by switching ON/OFF the small cells inside a Macro-cell without risking the bit rate received by the users. However, the resource allocation issues involved, e.g. base station selection and user association are NP-hard in general. In this paper we propose a novel solution, applying machine learning to train two neural networks to predict which base stations are not critical and can start their sleeping mode and also predict the associations between the users and base stations giving a complete solution for optimizing the energy efficiency in wireless networks. The outcome will provide similar results to the mathematical optimization but saving 99% of the time spent.

The efficient scheduling of transmission links in a wireless network with a certain optimization objective and subject to the interference and network flow constraints plays a central role in wireless networking research. As an alternative to traditional mathematical analysis, data-driven learning methods have shown promise in solving difficult problems by extracting knowledge from experiences and inspired applications of machine learning in wireless networking. In this paper, we focus on tackling the fundamental scheduling issue in multi-hop wireless networks with machine learning, facing the great challenges of the involvement of non-differentiable operations and the consideration of variable network topologies. To address these issues, we propose a reinforcement learning-based method to solve a class of network flow problems under the protocol interference model. Learning from experience, the proposed approach develops a strategy to sequentially select optimum subsets of links to transmit simultaneously to maximize the system throughput without causing interference. The model structure is designed in a way that incorporates network topological information to allow a flexible number of network nodes, and allows non-differentiable decision operation to pass informative gradient information. Experiments with synthetic and real-world deployment data demonstrate that the proposed algorithm achieves close-to-optimum performance at a significantly reduced time cost.

It is well known that the Proof-of-Work (PoW) based blockchain scheme, first introduced in Bitcoin system, is not scalable, where the PoW implementation limits the transaction processing rate. In recent years, parallel chain techniques have been proposed to overcome this issue. Chainweb is one of the parallel chains to be studied in this paper with a focus on security related issues. It is worth noting that existing blockchain security studies mainly focus on traditional single-chain based protocols. There are not many security related studies on parallel blockchains. This paper for the first time reveals that selfish mining attack is possible on Chainweb blockchain, to the best of our knowledge. Specifically, we propose a selfish mining attack that exclusively mines blocks on a subset of parallel chains with the same block height and achieves gain through a proper withholding strategy. We develop a mathematical model to quantitatively evaluate the performance and demonstrate the effectiveness of the proposed attack. Our results show that the attacker can gain extra mining reward when his computational power is at least 38% of the total power in Chainweb network. Under the 50% computational power restriction, the attacker's extra gain increases monotonically with its computational power.

Wireless optimization involves repeatedly solving difficult optimization problems, and data-driven deep learning techniques have great promise to alleviate this issue through its pattern matching capability: past optimal solutions can be used as the training data in a supervised learning paradigm so that the neural network can generate an approximate solution using a fraction of the computational cost, due to its high representing power and parallel implementation. However, making this approach practical in networking scenarios requires careful, domain-specific consideration, currently lacking in similar works. In this paper, we use deep learning in a wireless network scheduling and routing to predict if subsets of the network links are going to be used, so that the effective problem scale is reduced. A real-world concern is the varying data importance: training samples are not equally important due to class imbalance or different label quality. To compensate for this fact, we develop an adaptive sample weighting scheme which dynamically weights the batch samples in the training process. In addition, we design a novel loss function that uses additional network-layer feature information to improve the solution quality. We also discuss a post-processing step that gives a good threshold value to balance the trade-off between prediction quality and problem scale reduction. By numerical simulations, we demonstrate that these measures improve both the prediction quality and scale reduction when training from data of varied importance.

Recent advances in communications technologies and the proliferation of connected devices have spawned a variety of information-centric Internet-of-Things (IoT) systems, where timely information updating is normally required. Age of Information (AoI) has recently been introduced to quantify the freshness of the knowledge the controller has about the remote information sources. With the development of IoT applications, it is becoming increasingly common that the application-level services, e.g., status updates, rely on the timely delivery of fresh information from a number of information sources. In this article, we study the application-oriented scheduling for optimizing information freshness in the presence of correlated information sources. To this end, we adopt the concept of Age of Correlated Information (AoCI) and formulate the scheduling problem as an episodic Markov decision process (MDP) problem with an application-oriented policy. Given the complexity of the above problem, we develop a learning-based approach that not only leverages the emerging deep reinforcement learning (DRL) techniques but also exploits diverse-domain knowledge. The numerical results show that the proposed approach achieves better performance in terms of AoCI, compared to some typical baseline methods.

Motivated by the increasingly urgent demands for delivering fresh information, the age-of-information (AoI) has recently been introduced as an important metric for evaluating the timeliness performance of information update systems and has shed light on a number of research studies. Nevertheless, the most common goal of the existing works does not characterize the value of information freshness from the users' perspective. In this paper, we introduce the concept of effective AoI (EAoI) to quantify the freshness of the information users utilize for decision-making. We consider a general request-response model, which captures both proactive information update and timely information delivery, for investigating the scheduling problem with respect to EAoI minimization. By decomposing the scheduling problem into multiple computationally tractable subproblems, we propose request-aware scheduling policies for static and dynamic request models, respectively. The numerical results show that serving users requests proactively can reduce time-average EAoI in both scenarios.

The emerging ad hoc clouds form a new cloud computing paradigm by leveraging untapped local computation and storage resources. An important application of ad hoc clouds is to outsource computational intensive problems to nearby cloud agents. Specifically, for the problem of solving a linear algebraic equation (LAE), an outsourcing client assigns each cloud agent a subproblem, and then all involved agents apply a consensus-based algorithm to obtain the correct solution of the LAE in an iterative and distributed manner. However, such a distributed collaboration paradigm suffers from cyber security threats that undermine the confidentiality of the outsourced problem and the integrity of the returned results. In this paper, we identify a number of such security threats in this process, and propose a secure outsourcing scheme which not only preserves the privacy of the LAE parameters and the final solution from the participating agents, but also guarantees the correctness of the final solution. We prove that the proposed scheme has low computation complexity at each agent, and is robust against the identified security attacks. Numerical and simulation results are presented to demonstrate the effectiveness of the proposed method.

With the explosive growth of mobile applications and high computation burden on each single device, more and more end users demand to offload expensive computing tasks to external sites via job offloading technologies. Due to the fluctuating nature of jobs from end users, traditional cloud computing paradigm, however, has difficulties in accommodating highly dynamic job requests and meeting heterogeneous user requirements. Locating close to mobile users, edge-clouds have the potential to complement the cloud computing platform by acting as an efficient spot to perform users' deadline-sensitive tasks. In this paper, we study the resource allocation problem for accommodating deadline-sensitive jobs in edge-cloud system. We formulate a revenue maximization problem that captures the SLA-oriented property of job execution, and propose an efficient online multi-resource allocation algorithm that achieves low competitive ratio with moderate resource augmentation.

Mobile crowd sensing presents a new sensing paradigm which allows an individual participant with mobile device perform sensing tasks and activities of professional organizations. Privacy is one major concern in the mobile crowd sensing application. In this paper, we first present a fog-assisted mobile crowd sensing architecture, then we propose two privacy-preserving crowd sensing schemes for two categories of crowd sensing applications. The first scheme is based on an additive homomorphic encryption algorithm and allows the service subscriber collect the statistical data without revealing the individual data from each participant. The second scheme is based on a bitwise-XOR homomorphic encryption algorithm and allows the service subscriber collect the accurate data without know which data is from which participant. The proposed schemes can achieve κ-anonymity privacy level for each participant. We also give the performance analysis of the proposed schemes.

Fog computing is a promising architectural pattern to reduce the amount of data that is transferred to the cloud for processing and analysis. In this paper, we study fog-assisted data streaming scenario in which fog nodes at the network edge share their spare resources to help pre-process raw data of applications hosted in the cloud. A distributed resource sharing scheme is presented where the software defined network (SDN) controller dynamically adjusts the volume of application data that will be directed to fog nodes for pre-processing. The SDN controller makes decisions by coordinating fog nodes and cloud platform to collaboratively solve a social welfare maximization problem. Based on a hybrid alternating direction method of multipliers (H-ADMM) algorithm, computation burden for solving the optimization problem is fully distributed to fog nodes, cloud platform and SDN controller, where local variables of fog nodes are updated in parallel. With proper design of message exchange pattern, the communication overhead of the coordination to SDN controller grows smoothly with increasing number of participating fog nodes.

This paper studies the charging problem of plug-in hybrid electric vehicles (PHEVs) with a game theoretic approach. The interplay between PHEVs and smart micro-grid charging stations are modeled as a multi-leader-multi-flower Stackelberg game. In the game, each PHEV needs to select a charging station for maximum utility given the charging prices from each station; and each charging station needs to adjust its charging price for improved utility based on the charging requests received. An important issue being considered in this paper is that the actual cost for charging into a target energy level depends on the travel distance and traffic conditions between the requesting PHEV and the finally selected charging station. In this paper, we adopt a location aware approach to explicitly incorporate the location- related cost into the utility function in the game model. Note that the distance information, traffic information along the roads, and communications between PHEVs and charging stations are to be obtained or enabled by vehicular ad hoc networks (VANET). We develop the algorithms for charging station selection and price adjustment. Simulation results are presented to demonstrate that the utility improvement of the location-aware model over the location-blind model.

The proliferation of smart phones enables ubiquitous Mobile Ad-hoc Networks (MANETs) where mobile devices communicate with peers over a wireless channel in an ad hoc mode. In this paper, we introduce a novel method to achieve multi-hop communication among open-source, non-rooted Android devices using Wi-Fi Direct Technology, also known as Wi-Fi Peer-to-Peer (P2P). Then we implement a proactive routing protocol in an MANET using multiple off-the-shelf smart phones to enable efficient message delivery over a multi-hop MANET.

Wireless personal area network (WPAN) is small-ranged network centered at an individual for interconnecting personal devices. For such a network, the bootstrapping mechanism with which the devices establish a secure group key is of critical importance. Most existing bootstrapping mechanisms require out-of-band channels and involve human interactions for authentication. In this paper, we aim to develop a fully automated bootstrapping mechanism with only in-band channels with approvable security. Toward this end, we designed an integrity-guaranteed message (IGM) structure, a self-authenticated key agreement protocol, and a prescheduling mechanism in allusion to the IEEE 802.15.4 standard for WPANs. The IGM structure guarantees that an adversary cannot modify the IGM message without being detected, thus protects the message integrity without the requirement of shared secrets between the sender and the receiver devices. The proposed self-authenticated key agreement protocol utilizes the IGM's integrity guaranteed property, works together with the prescheduling mechanism to achieve message self-authentication, thus protecting the secure bootstrapping process from the node impersonation attack and the man-in-the-middle attack without leveraging any out-of-band channels. We analyze the security performance of the proposed schemes, and show that they can be seamless interoperative with the existing IEEE 802.15.4 standard.

With the rapid growth of smartphone and tablet users, Device-to-Device (D2D) communications have become an attractive solution for enhancing the performance of traditional cellular networks. However, relevant security issues involved in D2D communications have not been addressed yet. In this paper, we investigate the security requirements and challenges for D2D communications, and present a secure and efficient key agreement protocol, which enables two mobile devices to establish a shared secret key for D2D communications without prior knowledge. Our approach is based on the Diffie-Hellman key agreement protocol and commitment schemes. Compared to previous work, our proposed protocol introduces less communication and computation overhead. We present the design details and security analysis of the proposed protocol. We also integrate our proposed protocol into the existing Wi-Fi Direct protocol, and implement it using Android smartphones.