Click on the DOI link to access this article at the publishers website (may not be free).

With the growing demand for multimedia services, Dynamic Adaptive Streaming over HTTP (DASH) has become a key solution for delivering high-quality video content. In this work, we consider an Edge-DASH scenario and formulate a joint optimization problem that involves four critical aspects: bitrate allocation, user-to-server assignment, caching, and bandwidth allocation. Due to the complexity of the joint problem, we decompose it into sub-problems and address them separately. To solve the resulting sub-problems, we employ deep reinforcement learning, specifically the Deep Deterministic Policy Gradient (DDPG) method, for three of them, and develop a heuristic solution for the fourth. Simulation results demonstrate that our approach enhances performance across multiple metrics, including improved video delivery, reduced buffer underflow and overflow, and more efficient caching, which collectively enable greater utilization of edge resources for streaming. Moreover, we evaluated inference latency across edge and cloud hardware, confirming sub- to few-millisecond performance suitable for real-time deployment. This showcases the benefits of combining learning-based and heuristic techniques to meet the growing demand for adaptive video streaming in edge computing environments.

In additive manufacturing (AM), factories can quickly adjust their production to meet dynamic object demand without losing profitability, which allows them to implement less complex and more efficient supply chains. To facilitate the adoption of simplified supply chains, we propose the AM Cloud, where a set of micro-manufacturers pool their resources and offer them in an on-demand and pay-per-use basis. However, managing the AM Cloud requires setting a price for objects that incentivizes demand while guaranteeing that the manufacturers can earn a profit. It also requires finding control decisions that minimize operating costs under uncertain object demand. To address these challenges, we first propose a randomized multiobject auction that allocates printing area to the winning buyers and sets prices. We then design an online dynamic control algorithm that finds optimal decisions for the manufacturers to fulfill the object production orders from the auction winners without detailed knowledge of the system statistics. We show that the auction mechanism is truthful, individually rational, and guarantees that the manufacturers can make a profit. We also show that the solution obtained by our dynamic control algorithm is within a tight bound of the optimal one and can fulfill orders within a buyer-defined deadline.

Phishing emails are one of the most common and effective tools that cybercriminals use to gain access to an organization’s network or personal information. To detect these attacks, email service providers use a variety of tools and indicators, such as the URLs that attackers include in their email messages. However, cybercriminals are able to bypass these detection techniques by omitting URLs in their messages and instead engaging victims in a conversation to advance their attacks. In this paper, we investigate the performance of convolutional neural network (CNN) models that identify phishing attacks by analyzing only the text in the email messages. The models take as input an embedding of the text in the email’s body and output a probability indicating the likelihood that the message is malicious. We evaluate several CNN architectures using real-world phishing emails and find that the best performing one can identify phishing attacks with an accuracy of 98.139%, recall of 98.125%, and precision of 98.269%.

We are now in the big data era. Due to the emerging various systems and applications, such as the Internet of Things, cyber-physical systems, smart cities, smart healthcare, we are able to collect more data than ever before. On the other hand, it makes it very difficult to analyze such massive data in order to advance our science and engineering fields. We note that QR and LU factorizations are two of the most fundamental mathematical tools for data analysis. However, conducting QR or LU factorization of an m×n matrix requires computational complexity ofO(m2n). This incurs a formidable challenge in efficiently analyzing large-scale data sets by normal users or small companies on traditional resource-limited computers. To overcome this limitation, industry and academia propose to employ cloud computing that can offer abundant computing resources. This, however, obviously raises security concerns and hence a lot of users are reluctant to reveal their data to the cloud. To this end, we propose two secure outsourcing algorithms for efficiently performing large-scale QR and LU factorizations, respectively. We implement the proposed algorithms on the Amazon Elastic Compute Cloud (EC2) platform and a laptop. The experiment results show significant time saving for the user.

In a mobile edge computing (MEC) network, mobile devices, also called edge clients, offload their computations to multiple edge servers that provide additional computing resources. Since the edge servers are placed at the network edge, e.g., cell-phone towers, transmission delays between edge servers and edge clients are shorter compared to those of cloud computing. In addition, edge clients can offload their tasks to other nearby edge clients with available computing resources by exploiting the Fog Computing (FC) paradigm. A major challenge in MEC and FC networks is to assign the tasks from edge clients to edge servers, as well as to other edge clients, in such a way that their tasks are completed with minimum energy consumption and minimum processing delay. In this paper, we model task offloading in MEC as a constrained multi-objective optimization problem (CMOP) that minimizes both the energy consumption and task processing delay of the mobile devices. To solve the CMOP, we design an evolutionary algorithm that can efficiently find a representative sample of the best trade-offs between energy consumption and task processing delay, i.e., the Pareto-optimal front. Compared to existing approaches for task offloading in MEC, we see that our approach finds offloading decisions with lower energy consumption and task processing delay.

Big data has become a key basis of innovation and intelligence, potentially making our lives more convenient and bringing new opportunities to the modern society. Towards this goal, a critical underlying task is to solve a series of large-scale fundamental problems. Conducting such large-scale data analytics in a timely manner requires a large amount of computing resources, which may not be available for individuals and small companies in practice. By outsourcing their computations to the cloud, clients can solve such problems in a cost-effective way. However, confidential data stored at the cloud is vulnerable to cyber attacks, and thus needs to be protected. Previous works employ cryptographic techniques like homomorphic encryption, which significantly increase the computational complexity of solving a large-scale problem at the cloud and is impractical for big data applications. For the first time in the literature, we present an efficient secure outsourcing scheme for convex separable programming problems (CSPs). In particular, we first develop efficient matrix and vector transformation schemes only based on arithmetic operations that are computationally indistinguishable both in value and in structure under a chosen-plaintext attack (CPA). Then, we design a secure outsourcing scheme in which the client and the cloud collaboratively solve the transformed problems. The client can efficiently verify the correctness of returned results to prevent any malicious behavior of the cloud. Theoretical correctness and privacy analysis together show that the proposed scheme obtains optimal results and that the cloud cannot learn private information from the client's concealed data. We conduct extensive simulations on Amazon Elastic Cloud Computing (EC2) platform and find that our proposed scheme provides significant time savings to the clients.

Due to the large computational cost of data classification using deep learning, resource-limited devices, e.g., smart phones, PCs, etc., offload their classification tasks to a cloud server, which offers extensive hardware resources. Unfortunately, since the cloud is an untrusted third-party, users may be reluctant to share their private data with the cloud for data classification. Differential privacy has been proposed as a way of securely classifying data at the cloud using deep learning. In this approach, users conceal their data before uploading it to the cloud using a local obfuscation deep learning model, which is based on a data classification model hosted by the cloud. However, as the obfuscation model assumes that the pre-trained model at the cloud is static, it leads to significant performance degradation under realistic classification models that are constantly being updated. In this paper, we investigate the performance of differentially-private data classification under a dynamic pre-trained model, and a constant obfuscation model. We find that the classification performance decreases as the pre-trained model evolves. We then investigate the classification performance under an obfuscation model that is updated alongside the pre-trained model. We find that with a modest computational effort the obfuscation model can be updated to significantly improve the classification performance. under a dynamic pre-trained model.

Tensor decompositions are very powerful tools for analyzing multi-dimensional multi-modal data. Particularly, CP tensor decomposition is one of the most fundamental tensor decomposition models. However, it is usually computationally expensive to conduct CP tensor decompositions on a large-scale tensor by common algorithms like alternative least squares (ALS). To address this issue, one widely recognized solution is to adopt cloud computing. However, this raises privacy concerns due to the private information carried by a tensor. Previous algorithms for privacy-preserving outsourcing of tensor decompositions and other related computations require heavy communication cost. In this paper, we first develop an efficient tensor transformation scheme to protect the private information carried by elements' values of a tensor. Then we design a privacy-preserving outsourcing algorithm for ALS based CP tensor decompositions. We implement our proposed algorithm on a laptop and Amazon EC2 cloud and offer experiment results to show the sianificant computing time-savings.

Manufacturing systems are rapidly adopting the Internet of Things (IoT) to improve their efficiency and productivity. The IoT equips manufacturing systems with sensing, computing and communications capabilities, which enable real-time monitoring and control of increasingly complex and geographically distributed factory floors. However, the increased used of computer networks in IoT-enabled manufacturing introduces cyber-vulnerabilities that can be exploited by sophisticated adversaries to sabotage manufacturing operations. A particularly serious cyberattack against manufacturing systems is the defect injection (DI) attack. In a DI attack, a compromised machine fabricates objects with deformed geometry, weak material composition, abnormal dimensions, etc., which pose a great risk to safety-critical applications. In this paper, we develop a method to identify compromised machines that launch DI attacks against smart manufacturing systems. Specifically, we first propose a DI attack localization (DIAL) algorithm that uses machines' energy consumption and voltage measurements to identify compromised machines in the system. Our proposed approach only requires modest hardware resources and can be used in large-scale systems. We implement our DIAL algorithm on a real-world advanced manufacturing testbed, and observe that it can successfully locate the compromised machines with a high detection rate.

Solving large-scale sparse linear systems of equations (SLSEs) is one of the most common and fundamental problems in big data, but it is very challenging for resource-limited users. Cloud computing has been proposed as a timely, efficient, and cost-effective way of solving such expensive computing tasks. Nevertheless, one critical concern in cloud computing is data privacy. Specifically, clients’ SLSEs usually contain private information that should remain hidden from the cloud for ethical, legal, or security reasons. Many previous works on secure outsourcing of linear systems of equations (LSEs) have high computational complexity, and do not exploit the sparsity in the LSEs. More importantly, they share a common serious problem, i.e., a huge number of memory I/O operations. This problem has been largely neglected in the past, but in fact is of particular importance and may eventually render those outsourcing schemes impractical. In this paper, we develop an efficient and practical secure outsourcing algorithm for solving large-scale SLSEs, which has low computational and memory I/O complexities and can protect clients’ privacy well. We implement our algorithm on Amazon Elastic Compute Cloud, and find that the proposed algorithm offers significant time savings for the client (up to 74 percent) compared to previous algorithms.

Electric power systems are critical infrastructure and are vulnerable to contingencies including natural disasters, system errors, malicious attacks, etc. These contingencies can affect the world's economy and cause great inconvenience to our daily lives. Therefore, security of power systems has received enormous attention for decades. Recently, the development of the Internet of Things (IoT) enables power systems to support various network functions throughout the generation, transmission, distribution, and consumption of energy with IoT devices (such as sensors, smart meters, etc.). On the other hand, it also incurs many more security threats. Cascading failures, one of the most serious problems in power systems, can result in catastrophic impacts such as massive blackouts. More importantly, it can be taken advantage by malicious attackers to launch physical or cyber attacks on the power system. In this paper, we propose and investigate cascading failure attacks (CFAs) from a stochastic game perspective. In particular, we formulate a zerosum stochastic attack/defense game for CFAs while considering the attack/defense costs, budget constraints, diverse load shedding costs, and dynamic states in the system. Then, we develop a Q-CFA learning algorithm that works efficiently in power systems without any a priori information. We also formally prove that the convergence of the proposed algorithm achieves a Nash equilibrium. Simulation results validate the efficacy and efficiency of the proposed scheme by comparisons with other state-of-the-art approaches.

The massive amount of data that is being collected by today's society has the potential to advance scientific knowledge and boost innovations. However, people often lack sufficient computing resources to analyze their large-scale data in a cost-effective and timely way. Cloud computing offers access to vast computing resources on an on-demand and pay-per-use basis, which is a practical way for people to analyze their huge data sets. However, since their data contain sensitive information that needs to be kept secret for ethical, security, or legal reasons, many people are reluctant to adopt cloud computing. For the first time in the literature, we propose a secure outsourcing algorithm for large-scale quadratic programs (QPs), which is one of the most fundamental problems in data analysis. Specifically, based on simple linear algebra operations, we design a low-complexity QP transformation that protects the private data in a QP. We show that the transformed QP is computationally indistinguishable under a chosen plaintext attack (CPA), i.e., CPA-secure. We then develop a parallel algorithm to solve the transformed QP at the cloud, and efficiently find the solution to the original QP at the user. We implement the proposed algorithm on the Amazon Elastic Compute Cloud (EC2) and a laptop. We find that our proposed algorithm offers significant time savings for the user and is scalable to the size of the QP.

Solving large-scale linear systems of equations (LSEs) is one of the most common and fundamental problems in big data. But such problems are often too expensive to solve for resource-limited users. Cloud computing has been proposed as a timely, efficient, and cost-effective way of solving such computing tasks. Nevertheless, one critical concern in cloud computing is data privacy. To be more prominent, in many cases, clients's LSEs contain private data that should remain hidden from the cloud for ethical, legal, or security reasons. Many previous works on secure outsourcing of LSEs have high computational complexity. More importantly, they share a common serious problem, i.e., a huge number of external memory I/O operations. This problem has been largely neglected in the past, but in fact is of particular importance and may eventually render those outsourcing schemes impractical. In this paper, we develop an efficient and practical secure outsourcing algorithm for solving large-scale LSEs, which has both low computational complexity and low memory I/O complexity and can protect clients' privacy well. We implement our algorithm on a real-world cloud server and a laptop. We find that the proposed algorithm offers significant time savings for the client (up to 65%) compared to previous algorithms.

Due to the emerging various data services, current cellular networks have been experiencing a surge of data traffic and are already overloaded; thus, they are not able to meet the ever exploding traffic demand. In this study, we first introduce a multi-radio multi-channel multi-hop cognitive cellular network (M$^3$C$^2$N) architecture to enhance network throughput. Under the proposed architecture, we then investigate the minimum length scheduling problem by exploring joint frequency allocation, link scheduling, and routing. In particular, we first formulate a maximal independent set based joint scheduling and routing optimization problem called original optimization problem (OOP). It is a mixed integer non-linear programming (MINLP) and generally NP-hard problem. Then, employing a column generation based approach, we develop an$\epsilon$-bounded approximation algorithm which can obtain an$\epsilon$-bounded approximate result of OOP. Noticeably, in fact we do not need to find the maximal independent sets in the proposed algorithm, which are usually assumed to be given in previous works although finding all of them is NP-complete. We also revisit the minimum length scheduling problem by considering uncertain channel availability. Simulation results show that we can efficiently find the$\epsilon$-bounded approximate results and the optimal result as well, i.e., when$\epsilon =0\%$in the algorithm.

In ad hoc networks, selfish nodes deviating from the standard MAC (Medium Access Control) protocol can significantly degrade normal nodes' performance and are usually difficult to detect. In this paper, we propose detection and defense schemes to identify and defend against MAC-layer selfish misbehavior, respectively, in IEEE 802.11 multi-hop ad hoc networks. Specifically, the non-deterministic nature of the IEEE 802.11 MAC protocol imposes great challenges to distinguishing selfish nodes from well-behaved nodes. Most traditional selfish misbehavior detection approaches are for wireless local area networks (WLANs) only. They either rely on a large amount of historical data to perform statistical detection, or employ throughput or delay models that are only valid in WLANs for detection. In contrast, we propose a realtime selfish misbehavior detection scheme for multi-hop ad hoc networks. It requires only several samples, and hence is more efficient and can adapt to channel dynamics more quickly. Then, based on the proposed detection scheme, we design three selfish misbehavior defense schemes against three typical kinds of smart selfish nodes. We find that the smart selfish nodes cannot degrade normal nodes' performance much without getting detected. Extensive simulation results are finally presented to validate the proposed detection and defense schemes.

The explosive growth of Online Social Networks (OSNs) over the past few years has redefined the way people interact with existing friends and especially make new friends. Some works propose to let people become friends if they have similar profile attributes. However, profile matching involves an inherent privacy risk of exposing private profile information to strangers in the cyberspace. The existing solutions to the problem attempt to protect users’ privacy by privately computing the intersection or intersection cardinality of the profile attribute sets of two users. These schemes have some limitations and can still reveal users’ privacy. In this paper, we leverage community structures to redefine the OSN model and propose a realistic asymmetric social proximity measure between two users. Then, based on the proposed asymmetric social proximity, we design three private matching protocols, which provide different privacy levels and can protect users’ privacy better than the previous works. We also analyze the computation and communication cost of these protocols. Finally, we validate our proposed asymmetric proximity measure using real social network data and conduct extensive simulations to evaluate the performance of the proposed protocols in terms of computation cost, communication cost, total running time, and energy consumption. The results show the efficacy of our proposed proximity measure and better performance of our protocols over the state-of-the-art protocols.

With the ever increasing user adoption of mobile devices like smart phones and tablets, the cellular service providers' energy consumption and cost are fast-growing and have received tremendous attention. How to effectively reduce the energy cost of cellular networks and achieve green communications while satisfying cellular users' rocketing traffic demands has become an urgent and challenging problem. In this paper, we investigate the minimization of the long-term time-averaged expected energy cost of a cellular service provider while guaranteeing the strong stability of the network. We first formulate an offline optimization problem with a joint consideration of flow routing, link scheduling, and energy (i.e., renewable energy resource, energy storage unit, etc.) constraints. Since the formulated problem is a time-coupling stochastic Mixed-Integer Non-Linear Programming (MINLP) problem, it is prohibitively expensive to solve. Then, we reformulate the problem by employing Lyapunov optimization theory. A decomposition based algorithm is developed to solve the problem, which is proved to guarantee the network strong stability. Both the lower and upper bounds on the optimal result of the original problem are derived and proven. Simulation results demonstrate that the obtained lower and upper bounds are very tight, and that the proposed scheme results in noticeable energy cost savings.

The proliferation of mobile devices has driven the mobile marketing to surge in the past few years. Emerging as a new type of mobile marketing, mobile location-based services (MLBSs) have attracted intense attention recently. Unfortunately, current MLBSs have a lot of limitations and raise many concerns, especially about system security and users' privacy. In this paper, we propose a new location-based rewarding system, called LocaWard, where mobile users can collect location-based tokens from token distributors, and then redeem their gathered tokens at token collectors for beneficial rewards. Tokens act as virtual currency. The token distributors and collectors can be any commercial entities or merchants that wish to attract customers through such a promotion system, such as stores, restaurants, and car rental companies. We develop a security and privacy aware location-based rewarding protocol for the LocaWard system, and prove the completeness and soundness of the protocol. Moreover, we show that the system is resilient to various attacks and mobile users' privacy can be well protected in the meantime. We finally implement the system and conduct extensive experiments to validate the system efficiency in terms of computation, communication, energy consumption, and storage costs.

With great advances in mobile devices, e.g., smart phones and tablets, location-based services (LBSs) have recently emerged as a very popular application in mobile networks. However, since LBS service providers require users to report their location information, how to preserve users' location privacy is one of the most challenging problems in LBSs. Most existing approaches either cannot fully protect users' location privacy, or cannot provide accurate LBSs. Many of them also need the help of a trusted third-party, which may not always be available. In this paper, we propose a geometric approach, called n-CD, to provide realtime accurate LBSs while preserving users' location privacy without involving any third-party. Specifically, we first divide a user's region of interest (ROI), which is a disk centered at the user's location, into n equal sectors. Then, we generate n concealing disks (CDs), one for each sector, one by one to collaboratively and fully cover each of the n sectors. We call the area covered by the n CDs the concealing space, which fully contains the user's ROI. After rotating the concealing space with respect to the user's location, we send the rotated centers of the n CDs along with their radii to the service provider, instead of the user's real location and his/her ROI. To investigate the performance of n-CD, we theoretically analyze its privacy level and concealing cost. Extensive simulations are finally conducted to evaluate the efficacy and efficiency of the proposed schemes.

In the U.S., energy theft causes six billion dollar losses to utility companies (UCs) every year. With the smart grid being proposed to modernize current power grids, energy theft may become an even more serious problem since the “smart meters” used in smart grids are vulnerable to more types of attacks compared to traditional mechanical meters. Therefore, it is important to develop efficient and reliable methods to identify illegal users who are committing energy theft. Although some schemes have been proposed for the UCs to detect energy theft in power grids, they all require the users to send their private information, e.g., load files or meter readings at certain times, to the UCs which invades users' privacy and raises serious concerns about privacy, safety, etc. As far as we know, we are the first to investigate the energy theft detection problem considering users' privacy issues. In this paper, we propose to solve in a distributed fashion a linear system of equations (LSE) for the users' “honesty coefficients”, which indicate the users are honest when equal to 1 and are fraudulent when larger than 1. In particular, we develop two distributed privacy-preserving energy theft detection algorithms based on LU decomposition, called LUD and LUPD, respectively, which can identify fraudulent users without invading any user's privacy. Compared to LUD, LUPD requires higher execution time but is stable even in large-size systems. Moreover, the LUD and LUPD algorithms are proposed in the case that users commit energy theft at a constant rate, i.e., with constant honesty coefficients. We also propose adaptive LUD/LUPD algorithms to account for the scenarios where the users have variable honesty coefficients. Extensive simulations are carried out and the results show that the proposed algorithms can efficiently and successfully identify the fraudulent users in the system.