Tearing the anterior cruciate ligament requires repair and rehabilitation to restore lower limb functionality fully. This study outlines a method for monitoring rehabilitation after knee surgery by analyzing footstep sounds and applying deep learning techniques. The process involves examining gait sounds during the initial four weeks of rehabilitation. The suggested system, StrideSense, recognizes walking sounds, enabling seamless and ongoing monitoring of patients’ gait rehabilitation. The research proposes a novel method for event detection by analyzing walking patterns using dynamic time-warping and sequential footstep duration. It leverages physiological data like gait sound and energy descriptors to develop a deep-learning model for gait improvement assessment, evaluated by a physiotherapist through lower extremity functional scores (LEFS). The suggested model, AtdNet, which utilizes Densenet169 and attention mechanisms, achieves 96% accuracy in classifying walking on various post-surgical days. It also predicts LEFS with a mean absolute error of 4.63%. A deeper analysis of bone-conducted footstep sounds enriched the acoustic sensing method. We assessed the suggested acoustic model alongside existing methods, showing that rehabilitation monitoring driven by acoustics outperforms traditional clinic-based approach assessments. Future efforts will focus on validating the model with a larger dataset and integrating it into smart homes.

A data publishing deal conducted with anonymous microdata can preserve the privacy of people. However, anonymizing data with multiple records of an individual (1:M dataset) is still a challenging problem. After anonymizing the 1:M microdata, the vertical correlation can be exploited to launch privacy attacks. In this paper, a novel privacy preserving model$l_{c}, l_{s}$-ANGEL is proposed. To validate the new model, two privacy attacks are presented, namely, a Vertical correlation attack ($V_{c0}$) and a Vulnerable sensitive attribute attack ($V_{sa}$) on 1:M datasets, which breach the privacy of individuals. Furthermore, the proposed model is examined through High-Level Petri Nets (HLPNs). Our experiments on three real-world datasets;“INFORMS”,“YOUTUBE”, and “IMDb” demonstrate that the proposed model outperforms the state-of-the-art models. Our practices and lessons learned in this work can direct future concrete steps towards Multiple Sensitive Attributes, where we can expand the proposed model to dynamic datasets.

Vehicle-to-grid (V2G) energy trading based on distributed ledger technologies (DLT), such as blockchains, has attracted much attention due to its promising features, including ease of deployment, decentralization, transparency, and security. However, existing DLT-based models do not support microtransactions due to the low value of such transactions relative to the incentives offered to transaction verifiers. To address this issue, we propose an IOTA DLT-based efficient and secure energy trading model for V2G networks, where electric vehicles (EVs) and grids negotiate energy prices in an off-chain manner. The proposed model utilizes a privacy-preserving protocol to prevent real-time tracking of EV locations. We develop a Stackelberg game model to represent the interactions between the EVs and grids, from which we derive a pricing scheme and propose a deposit mechanism to prevent fake energy trading between the EVs and grids. Extensive simulations demonstrate that our proposed scheme outperforms existing V2G energy trading mechanisms regarding transaction efficiency, provides enhanced EV privacy, and improves resilience against fake energy trading. Offering robust computational performance and addressing computational complexity (time, space, and message), our model presents a comprehensive V2G energy trading solution, balancing efficiency, security, and privacy.

The increasing prevalence of blockchain-based cryptocurrencies as a payment instrument in the past decade and the rewards earned by the cryptominers has resulted in a new class of cyber attacks,cryptojacking, which involves unauthorized mining of cryptocurrencies on someone's system. Spotting cryptojacking is difficult in many cases, since the relevant software tries to disguise its presence to evade detection, by mimicking benign software such as compression applications by performing similar bitwise, cryptographic, and encryption operations. In this paper, we propose the processing of assembly code—a fundamental and platform-independent programming language—as a natural language using deep learning for profiling applications, which we callDeepCodeProfiler (DeCode Pro). Our proposed solution leverages the immutable step of any cyber attack: the deployment of instructions in system memory to carry out the attack. Through extensive experimentation with different neural network architectures in the profiling stage, we show that DeCode Pro is highly effective in the detection of evasive cryptojacking attacks and achieves low false positive and false negative rates. We also show that the model achieves high classification accuracy even with limited training data, which can considerably reduce the computing resources required for training and retraining the deep learning model.

Moving target defenses (MTDs) are widely used as an active defense strategy for thwarting cyberattacks on cyber-physical systems by increasing diversity of software and network paths. Recently, machine Learning (ML) and deep Learning (DL) models have been demonstrated to defeat some of the cyber defenses by learning attack detection patterns and defense strategies. It raises concerns about the susceptibility of MTD to ML and DL methods. In this article, we analyze the effectiveness of ML and DL models when it comes to deciphering MTD methods and ultimately evade MTD-based protections in real-time systems. Specifically, we consider a MTD algorithm that periodically randomizes address assignments within the MIL-STD-1553 protocol—a military standard serial data bus. Two ML and DL-based tasks are performed on MIL-STD-1553 protocol to measure the effectiveness of the learning models in deciphering the MTD algorithm: 1) determining whether there is an address assignments change i.e., whether the given system employs a MTD protocol and if it does 2) predicting the future address assignments. The supervised learning models (random forest and k-nearest neighbors) effectively detected the address assignment changes and classified whether the given system is equipped with a specified MTD protocol. On the other hand, the unsupervised learning model (K-means) was significantly less effective. The DL model (long short-term memory) was able to predict the future addresses with varied effectiveness based on MTD algorithm's settings.

Word embeddings have become the de-facto tool for representing text in natural language processing (NLP) tasks, as they can capture semantic and syntactic relations, unlike their precedents such as Bag-of-Words. Although word embeddings have been employed in various studies in recent years and proven to be effective in many NLP tasks, they are still immature for sentiment analysis, as they suffer from insufficient sentiment information. General word embedding models pre-trained on large corpora with methods such as Word2Vec or GloVe achieve limited success in domain-specific NLP tasks. On the other hand, training domain-specific word embeddings from scratch requires a high amount of data and computation power. In this work, we target both shortcomings of pre-trained word embeddings to boost the performance of domain-specific sentiment analysis tasks. We propose a model that refines pre-trained word embeddings with context information and leverages the sentiment scores of sentences obtained from a lexicon-based method to further improve performance. Experiment results on two benchmark datasets show that the proposed method significantly increases the accuracy of sentiment classification.

The number of Internet of Things (IoT) devices and the use cases they aim to support have increased sharply in the past decade with the rapid developments in wireless networking infrastructures. Despite many advantages, the widespread use of IoT has also created a large attack surface frequently exploited by cyber criminals, requiring real-time, automated detection and mitigation of various attacks in the high-volume network traffic generated. Software-defined networking (SDN) and machine learning (ML) based intrusion detection are effective tools for providing quick response to various attacks in IoT networks, however the study of ML-based intrusion detection so far has been limited to performance studies on datasets that were created a long while ago and are not specific to SDN-based environments. In this paper we introduce a novel dataset for intrusion detection in IoT networks. The dataset comprises two parts modeling static and dynamic IoT networks and consists of 27.9 million and 30.2 million data records respectively, which contain cyber attacks of various types in addition to benign traffic. The dataset will be an important resource for intrusion detection research in SDN-managed IoT, which will be increasingly prevalent in the future networks of ubiquitous connectivity.

The Internet of Things (IoT) is now being used increasingly in transportation, healthcare, agriculture, smart home and city systems. IoT devices, the number of which is expected to reach 25 billion all over the world by 2021, are required to be deployed very fast, taking into account commercial pressures. This results in a very important layer, i.e. security, being either completely neglected or having significant shortcomings. Since IoT has a heterogeneous structure, there is a need for intrusion detection systems (IDSs) that take into account the specifics of an IoT system architecture, including the computing power limitations, variety of protocols and prevalence of zero-day attacks. In this paper, we describe ARTEMIS, an IDS for IoT, which processes data from IoT devices using machine learning to detect deviations from the normal behavior of the system and generates alerts in case of anomalies. We have implemented a prototype of the system using IoT devices subscribed to topics at an MQTT broker and provide experimental evaluation of the system under MQTT-related attacks.

e era of Internet of ings with billions of connected devices has created an ever larger surface for cyber attackers to exploit, which has resulted in the need for fast and accurate detection of those attacks.e developments in mobile computing, communications, and mass storage architectures in the past decade have brought about the phenomenon of big data, which involves unprecedented amounts of valuable data generated in various forms at a high speed.e ability to process these massive amounts of data in real time using big data analytics tools brings along many bene ts that could be utilized in cyber threat analysis systems.By making use of big data collected from networks, computers, sensors, and cloud systems, cyber threat analysts and intrusion detection/prevention systems can discover useful information in real time.is information can help detect system vulnerabilities and attacks that are becoming prevalent and develop security solutions accordingly.Big data analytics will be a must-have component of any e ective cyber security solution due to the need of fast processing of the high-velocity, high-volume data from various sources to discover anomalies and/or attack patterns as fast as possible to limit the vulnerability of the systems and increase their resilience.Even though many big data analytics tools have been developed in the past few years, their usage in the eld of cyber security warrants new approaches considering many aspects including (a) uni ed data representation, (b) zero-day attack detection, (c) data sharing across threat detection systems, (d) real time analysis, (e) sampling and dimensionality reduction, (f ) resource-constrained data processing, and (g) time series analysis for anomaly detection.is special issue has attracted original contributions that utilize and build big data analytics solutions for cyber

With advances in cloud computing and the emergence of service marketplaces, the popularity of composite services marks a paradigm shift from single-domain monolithic systems to cross-domain distributed services, which raises important privacy and security concerns. Access control becomes a challenge in such systems because authentication, authorization and data disclosure may take place across endpoints that are not known to clients. The clients lack options for specifying policies to control the sharing of their data and have to rely on service providers which offer limited selection of security and privacy preferences. This lack of awareness and loss of control over data sharing increases threats to a client's data and diminishes trust in these systems. We propose EPICS, an efficient and effective solution for enforcing security policies in composite Web services that protects data privacy throughout the service interaction lifecycle. The solution ensures that the data are distributed along with the client policies that dictate data access and an execution monitor that controls data disclosure. It empowers data owners with control of data disclosure decisions during interactions with remote services and reduces the risk of unauthorized access. The paper presents the design, implementation, and evaluation of the EPICS framework.

When clients interact with a cloud-based service, they expect certain levels of quality of service guarantees. These are expressed as security and privacy policies, interaction authorization policies, and service performance policies among others. The main security challenge in a cloud-based service environment, typically modeled using service-oriented architecture (SOA), is that it is difficult to trust all services in a service composition. In addition, the details of the services involved in an end-to-end service invocation chain are usually not exposed to the clients. The complexity of the SOA services and multi-tenancy in the cloud environment leads to a large attack surface. In this paper we propose a novel approach for end-to-end security and privacy in cloud-based service orchestrations, which uses a service activity monitor to audit activities of services in a domain. The service monitor intercepts interactions between a client and services, as well as among services, and provides a pluggable interface for different modules to analyze service interactions and make dynamic decisions based on security policies defined over the service domain. Experiments with a real-world service composition scenario demonstrate that the overhead of monitoring is acceptable for real-time operation of Web services.

Advances in cloud computing have made it a feasible and cost-effective solution to improve the resiliency of enterprise systems. However, the replication approach taken by cloud computing to provide resiliency leads to an increase in the number of ways an attacker can exploit or penetrate the systems. This calls for designing cloud systems that can accurately detect anomalies and dynamically adapt themselves to keep performing mission-critical functions even under attacks and failures. In this paper, we propose a self-adaptive resiliency approach for cloud enterprise systems that employs a live monitoring and moving target defense based approach to automatically detect deviations from normal behavior and reconfigure critical cloud processes through software-defined networking to mitigate attacks and reduce system downtime. The proposed solution is promising to present a unified framework for resilient cloud systems.

Consumer ratings are widely used in online marketplaces-helping vendors in assessing the quality of offerings and consumers in discovery and purchase decisions. To build trust in a marketplace, which has a direct impact on sales, an accurate assessment of ratings is essential in determining the quality of offerings. This paper proposes novel extensions to consumer Rating as a Service (RaaS)-a rating management service providing consumer rating functionality to a marketplace using hierarchical aggregation, which is a rating aggregation mechanism using hierarchical relationships of components to evaluate composite offerings. Contributions include the optimization of RaaS design for Web-scale, the integration of consumer credibility in hierarchical aggregation, and the application of hierarchical aggregation to existing independent atomic offerings. Various experiments are conducted to demonstrate the practicality of RaaS and correctness of hierarchical aggregation using real ratings from Amazon.com.

The rise of the mobile-cloud computing paradigm has enabled mobile devices with limited processing power and battery life to achieve complex tasks in real-time. While mobile-cloud computing is promising to overcome limitations of mobile devices for real-time computing needs, the reliance of existing models on strong assumptions such as the availability of a full clone of the application code and non-standard system environments in the cloud makes it harder to manage the performance of mobile-cloud computing based applications. Furthermore, offloading mobile computation to the cloud entails security risks associated with sending data and code to an untrusted platform and perfect security is hard to achieve due to the extra computational overhead introduced by complex mechanisms. In this paper, we present a dynamic computation-offloading model for mobile-cloud computing, based on autonomous agent-based application partitions. We propose a dynamic tamper-resistance approach for managing the security of offloaded computation, by augmenting agents with self-protection capability using a low-overhead introspection and integrity-preserving communication mechanism. Experiments with a real-world mobile application demonstrates the effectiveness of the approach for high-performance, tamper-resistant mobile-cloud computing.

The rise of the mobile-cloud computing paradigm in recent years has enabled mobile devices with processing power and battery life limitations to achieve complex tasks in real-time. While mobile-cloud computing is promising to overcome the limitations of mobile devices for real-time computing, the lack of frameworks compatible with standard technologies and techniques for dynamic performance estimation and program component relocation makes it harder to adopt mobile-cloud computing at large. Most of the available frameworks rely on strong assumptions such as the availability of a full clone of the application code and negligible execution time in the cloud. In this paper, we present a dynamic computation offloading model for mobile-cloud computing, based on autonomous agents. Our approach does not impose any requirements on the cloud platform other than providing isolated execution containers, and it alleviates the management burden of offloaded code by the mobile platform using stateful, autonomous application partitions. We also investigate the effects of different cloud runtime environment conditions on the performance of mobile-cloud computing, and present a simple and low-overhead dynamic make span estimation model integrated into autonomous agents to enhance them with self-performance evaluation in addition to self-cloning capabilities. The proposed performance profiling model is used in conjunction with a cloud resource optimization scheme to ensure optimal performance. Experiments with two mobile applications demonstrate the effectiveness of the proposed approach for high-performance mobile-cloud computing.

Development of authorization mechanisms for secure information access by a large community of users in an open environment is an important problem in the ever-growing Internet world. In this paper we propose a computational dynamic trust model for user authorization, rooted in findings from social science. Unlike most existing computational trust models, this model distinguishes trusting belief in integrity from that in competence in different contexts and accounts for subjectivity in the evaluation of a particular trustee by different trusters. Simulation studies were conducted to compare the performance of the proposed integrity belief model with other trust models from the literature for different user behavior patterns. Experiments show that the proposed model achieves higher performance than other models especially in predicting the behavior of unstable users.

A security feature is a customer-valued capability of software for mitigating a set of security threats. Incremental development of security features, using the Scrum method, often leads to developing ineffective features in addressing the threats they target due to factors such as incomplete security tests. This paper proposes the use of security assurance cases to maintain a global view of the security claims as the feature is being developed iteratively and a process that enables the incremental development of security features while ensuring the security requirements of the feature are fulfilled.

The agile software development approach makes developing secure software challenging. Existing approaches for extending the agile development process, which enables incremental and iterative software development, fall short of providing a method for efficiently ensuring the security of the software increments produced at the end of each iteration. This article (a) proposes a method for security reassurance of software increments and demonstrates it through a simple case study, (b) integrates security engineering activities into the agile software development process and uses the security reassurance method to ensure producing acceptably secure-by the business owner-software increments at the end of each iteration, and (c) discusses the compliance of the proposed method with the agile values and its ability to produce secure software increments.

Service-Oriented Architecture (SOA) is becoming a major paradigm for distributed application development in the recent explosion of Internet services and cloud computing. However, SOA introduces new security challenges not present in the single-hop client-server architectures due to the involvement of multiple service providers in a service request. The interactions of independent service domains in SOA could violate service policies or SLAs. In addition, users in SOA systems have no control on what happens in the chain of service invocations. Although the establishment of trust across all involved partners is required as a prerequisite to ensure secure interactions, still a new end-to-end security auditing mechanism is needed to verify the actual service invocations and its conformance to the expected service orchestration. In this paper, we provide an efficient solution for end-to-end security auditing in SOA. The proposed security architecture introduces two new components called taint analysis and trust broker in addition to taking advantages of WS-Security and WS-Trust standards. The interaction of these components maintains session auditing and dynamic trust among services. This solution is transparent to the services, which allows auditing of legacy services without modification. Moreover, we have implemented a prototype of the proposed approach and verified its effectiveness in a LAN setting and the Amazon EC2 cloud computing infrastructure.

Context-awareness is a critical aspect of safe navigation especially for the blind and visually impaired in unfamiliar environments. Existing mobile devices for context-aware navigation fall short in many cases due to their dependence on specific infrastructure requirements as well as having limited access to resources that could provide a wealth of contextual clues. In this paper, we propose a mobile-cloud collaborative approach for context-aware navigation by exploiting the computational power of resources made available by Cloud Computing providers as well as the wealth of location-specific resources available on the Internet. We propose an extensible system architecture that minimizes reliance on infrastructure, thus allowing for wide usability. We present a traffic light detector that we developed as an initial application component of the proposed system. We present preliminary results of experiments performed to test the appropriateness for the real-time nature of the application.

Entities (e.g., users, services) have to authenticate themselves to service providers (SPs) in order to use their services. An entity provides personally identifiable information (PII) that uniquely identifies it to an SP. In the traditional application-centric Identity Management (IDM) model, each application keeps trace of identities of the entities that use it. In cloud computing, entities may have multiple accounts associated with different SPs, or one SP. Sharing PIIs of the same entity across services along with associated attributes can lead to mapping of PIIs to the entity. We propose an entity-centric approach for IDM in the cloud. The approach is based on: (1) active bundles-each including a payload of PII, privacy policies and a virtual machine that enforces the policies and uses a set of protection mechanisms to protect themselves, (2) anonymous identification to mediate interactions between the entity and cloud services using entity's privacy policies. The main characteristics of the approach are: it is independent of third party, gives minimum information to the SP and provides ability to use identity data on untrusted hosts.

Recent research has shown that collective classification in relational data often exhibit significant performance gains over conventional approaches that classify instances individually. This is primarily due to the presence of autocorrelation in relational datasets, meaning that the class labels of related entities are correlated and inferences about one instance can be used to improve inferences about linked instances. Statistical relational learning techniques exploit relational autocorrelation by modeling global autocorrelation dependencies under the assumption that the level of autocorrelation is stationary throughout the dataset. To date, there has been no work examining the appropriateness of this stationarity assumption. In this paper, we examine two real-world datasets and show that there is significant variance in the autocorrelation dependencies throughout the relational data graphs. We develop a shrinkage technique for modeling this non-stationary autocorrelation and show that it achieves significant accuracy gains over competing techniques that model either local or global autocorrelation dependencies in isolation.