Search

The COVID-19 pandemic is an unprecedented global emergency. Clinicians and medical researchers are suddenly thrown into a situation where they need to keep up with the latest and best evidence for decision-making at work in order to save lives and develop solutions for COVID-19 treatments and preventions. However, a challenge is the overwhelming numbers of online publications with a wide range of quality. We explain a science gateway platform designed to help users to filter the overwhelming amount of literature efficiently (with speed) and effectively (with quality), to find answers to their scientific questions. It is equipped with a chatbot to assist users to overcome infodemic, low usability, and high learning curve. We argue that human-machine communication via a chatbot play a critical role in enabling the diffusion of innovations.

Understanding and controlling the self-assembly of vertically oriented carbon nanotube (CNT) forests is essential for realizing their potential in myriad applications. The governing process–structure–property mechanisms are poorly understood, and the processing parameter space is far too vast to exhaustively explore experimentally. We overcome these limitations by using a physics-based simulation as a high-throughput virtual laboratory and image-based machine learning to relate CNT forest synthesis attributes to their mechanical performance. Using CNTNet, our image-based deep learning classifier module trained with synthetic imagery, combinations of CNT diameter, density, and population growth rate classes were labeled with an accuracy of >91%. The CNTNet regression module predicted CNT forest stiffness and buckling load properties with a lower root-mean-square error than that of a regression predictor based on CNT physical parameters. These results demonstrate that image-based machine learning trained using only simulated imagery can distinguish subtle CNT forest morphological features to predict physical material properties with high accuracy. CNTNet paves the way to incorporate scanning electron microscope imagery for high-throughput material discovery.

New virtualization technologies allow Infrastructure Providers (InPs) to lease their resources to Application Service Providers (ASPs) for highly scalable delivery of cloud services to end-users. However, existing literature lacks knowledge on Quality of Experience (QoE)-oriented cloud service orchestration algorithms that can guide ASPs on how to plan their budget to enhance satisfactory QoE delivery to end-users. In contrast to the InP’s cloud service orchestration, the ASP’s orchestration should not rely on expensive infrastructure control mechanisms such as Software-Defined Networking (SDN), or require apriori knowledge on the number of services to be instantiated and their anticipated placement location within InP’s infrastructure. In this paper, we address this issue of delivering satisfactory user QoE by synergistically optimizing both ASP’s management and data planes . The optimization within the ASP management plane first maximizes Service Level Objective (SLO) coverage of users when application services are being deployed, and are not yet operational. The optimization of the ASP data plane then enhances satisfactory user QoE delivery when applications services are operational with real user access. Our evaluation of QoE-oriented algorithms using realistic numerical simulations, real-world cloud testbed experiments with actual users and ASP case studies show notably improved performance over existing cloud service orchestration solutions.

There is an increasing trend in cloud adoption of enterprise applications in, for example, manufacturing, healthcare, and finance. Such applications are routinely subject to targeted cyberattacks, which result in significant loss of sensitive data (e.g., due to data exfiltration in advanced persistent threats) or valuable utilities (e.g., due to resource the exfiltration of power in cryptojacking). There is a critical need to train highly skilled cybersecurity professionals, who are capable of defending against such targeted attacks. In this article, we present the design, development, and evaluation of the Mizzou Cyber Range, an online platform to learn basic/advanced cyber defense concepts and perform training exercises to engender the next-generation cybersecurity workforce. Mizzou Cyber Range features flexibility, scalability, portability, and extendability in delivering cyberattack/defense learning modules to students. We detail our “research-inspired learning” and “learn–apply–create” three-phase pedagogy methodologies in the development of four learning modules that include laboratory exercises and self-study activities using realistic cloud-based application testbeds. The learning modules allow students to gain skills in using latest technologies (e.g., elastic capacity provisioning, software-defined everything infrastructure) to implement sophisticated “attack defense by pretense” techniques. Students can also use the learning modules to understand the attacker–defender game in order to create disincentives (i.e., pretense initiation) that make the attacker's tasks more difficult, costly, time consuming, and uncertain. Lastly, we show the benefits of our Mizzou Cyber Range through the evaluation of student learning using auto-grading, rank assessments with peer standing, and monitoring of students’ performance via feedback from prelab evaluation surveys and postlab technical assessments.

Healthcare innovations are increasingly becoming reliant on high variety and standards-compliant (e.g., HIPAA, common data model) distributed data sets that enable predictive analytics. Consequently, health information systems need to be developed using cooperation and distributed trust principles to allow protected data sharing between multiple domains or entities (e.g., health data service providers, hospitals and research labs). In this paper, we present a novel health information sharing system viz., HonestChain that uses Blockchain technology to allow organizations to have incentive-based and trustworthy cooperation to either access or provide protected healthcare records. More specifically, we use a consortium Blockchain approach coupled with chatbot guided interfaces that allow data requesters to: (a) comply with data access standards, and (b) allow them to gain reputation in a consortium. We also propose a reputation scheme for creation and sustenance of the consortium with peers using Requester Reputation and Provider Reputation metrics. We evaluate HonestChain using Hyperledger Composer in a realistic simulation testbed on a public cloud infrastructure. Our results show that our HonestChain performs better than the state-of-the-art requester reputation schemes for data request handling, while choosing the most appropriate provider peers. We particularly show that HonestChain achieves a better tradeoff in metrics such as service time and request resubmission rate. Additionally, we also demonstrate the scalability of our consortium platform in terms of the Blockchain transaction times.

The rapid growth of COVID-19 publications has driven clinical researchers and healthcare professionals in pursuit to reduce the knowledge gap on reliable information for effective pandemic solutions. The manual task of retrieving high-quality publications based on the evidence pyramid levels, however, presents a major bottleneck in researchers’ workflows. In this paper, we propose an “evidence-based” recommender system namely, KnowCOVID-19 that utilizes an edge computing service to integrate recommender modules for data analytics using end-user thin-clients. The edge computing service features chatbot-based web interface that handles a given COVID-19 publication dataset using two recommender system modules: (i) evidence-based filtering that observes domain specific topics across the literature and classifies the filtered information according to a clinical category, and (ii) social filtering that allows diverse experts with similar objectives to collaborate via a “social plane” to jointly find answers to critical clinical questions to fight the pandemic. We compare the Domain-specific Topic Model (DSTM) used in our evidence-based filtering with state-of-the-art models considering the CORD-19 dataset (a COVID-19 publication archive) and show improved generalization effectiveness as well as knowledge pattern query effectiveness. In addition, we conduct a comparison study between a manual literature review process and the KnowCOVID-19 augmented process, and evaluate the benefits of our information retrieval techniques over important queries provided by COVID-19 clinical experts.

Federated multi-cloud resource allocation for data-intensive application workflows is generally performed based on performance or quality of service (i.e., QSpecs ) considerations. At the same time, end-to-end security requirements of these workflows across multiple domains are considered as an afterthought due to lack of standardized formalization methods. Consequently, diverse/heterogenous domain resource and security policies cause inter-conflicts between application's security and performance requirements that lead to sub-optimal resource allocations. In this paper, we present a joint performance and security-driven federated resource allocation scheme for data-intensive scientific applications. In order to aid joint resource brokering among multi-cloud domains with diverse/heterogenous security postures, we first define and characterize a data-intensive application's security specifications (i.e., SSpecs ). Then we describe an alignment technique inspired by Portunes Algebra to homogenize the various domain resource policies (i.e., RSpecs ) along an application's workflow lifecycle stages. Using such formalization and alignment, we propose a near optimal cost-aware joint QSpecs - SSpecs -driven, RSpecs -compliant resource allocation algorithm for multi-cloud computing resource domain/location selection as well as network path selection. We implement our security formalization, alignment, and allocation scheme as a framework, viz., “OnTimeURB” and validate it in a multi-cloud environment with exemplar data-intensive application workflows involving distributed computing and remote instrumentation use cases with different performance and security requirements.

Traditionally, Network Function Virtualization uses Service Function Chaining (SFC) to place service functions and chain them with corresponding flows allocation. With the advent of Edge computing and IoT, a reliable orchestration of latency-sensitive SFCs is needed to compose and maintain them in geo-distributed cloud infrastructures. However, the optimal SFC composition in this case becomes the NP-hard integer multi-commodity-chain flow (MCCF) problem that has no known approximation guarantees. In this paper, we first outline our novel practical and near optimal SFC composition scheme which is based on our novel metapath composite variable approach, admits end-to-end network QoS constraints (e.g., latency) and reaches 99% optimality on average in seconds for practically sized geo-distributed cloud infrastructures. We then propose a novel metapath-based SFC maintenance algorithm that guarantees a distributed control plane consistency without use of expensive consensus protocols. Using trace-driven simulations comprising of challenging disaster-incident conditions, we show that our solution composes twice as many SFCs and uses $\sim$ 10x less control messages than state-of-the-art methods. Finally, experimental evaluations of our SFC orchestration prototype deployed on a realistic cloud/edge computing testbed show significant speed-ups (up to 3.5x) for our case-study geo-distributed latency-sensitive object tracking pipeline w.r.t. its IP-based cloud computing alternative.

Nowadays the abundance of IoT devices has the potential of changing our lives dramatically, but brings new routing and traffic orchestration challenges for the next-generation Internet providers: core routers are already overwhelmed, see e.g, the routing table size growth problem. Although some researchers still argue whether or not the next-generation networks should feature scale-free properties, recent results have shown benefits of embedding such scale-free networks in a hyperbolic space of negative curvature. Specifically, this allows geometrically route packets by using only a local topology knowledge ( i . e . , with average O ∗ ( 1 ) space–time complexity) at no extra communication overhead ( i . e . , without routing protocols). To our knowledge, however, there is no Traffic Engineering (TE) protocol with the aforementioned properties that can be used in dynamic scale-free networks. In this paper, we propose the first to our knowledge REpulsive-BAsed Traffic Engineering (REBATE) protocol for dynamic scale-free networks. REBATE is built upon dual principles of the demand-aware TE and fundamentals properties of hyperbolic spaces. Using trace-driven numerical simulations, we then show how REBATE can reduce the maximum link utilization up to 25% when compared to a common geometric routing-based traffic steering. Although REBATE can perform worse than common demands-aware and oblivious TE approaches, we think that our work should pave the way for more efficient TE in the next-generation dynamic scale-free networks.

Virtual Learning Environments (VLEs) are spaces designed to educate student groups remotely via online platforms. Although traditional VLEs have shown promise in educating students, they offer limited immersion that overall diminishes learning effectiveness. In this paper, we describe vSocial, a cloud-based virtual reality learning environment (VRLE) system that can be deployed over high-speed networks using the High Fidelity “social VR” platform. vSocial provides flexible control of group learning content and compliance with established VLE standards with improved immersive user experience for both instructor(s) and students. For our vSocial development, we build upon the use case of an existing special education VLE viz., iSocial that trains youth with Autism Spectrum Disorder by implementing the Social Competence Intervention (SCI) curriculum. The vSocial can be used to: (a) implement multiple learning modules using wearable VR technologies, (b) integrate cognitive state sensing devices, and (c) organize learning session data securely using web applications hosted on cloud resources. Our experiment results show that the VR mode of content delivery in vSocial better stimulates the generalization of lessons to the real world than non-VR lessons, and provides improved immersion when compared to an equivalent desktop version. Further, usability study results show that users can successfully use the web application features in vSocial for group learning activities with ease-of-use and consistency.

With the increase of DDoS attacks, resource adaptation schemes need to be effective to protect critical cloud-hosted applications. Specifically, they need to be adaptable to attack behavior, and be dynamic in terms of resource utilization. In this paper, we propose an intelligent strategy for proactive and reactive application migration by leveraging the concept of ‘moving target defense’ (MTD). The novelty of our approach lies in: (a) stochastic proactive migration frequency minimization across heterogeneous cloud resources to optimize migration management overheads, (b) market-driven migration location selection during proactive migration to optimize resource utilization, cloud service providers (CSPs) cost and user quality of experience, and (c) fast converging cost-minimizing reactive migration coupled with a ‘false reality’ pretense to reduce the future attack success probability. We evaluate the effectiveness of our proposed MTD-based defense strategy using a Software-defined Networking (SDN) enabled GENI Cloud testbed for a “Just-in-time news articles and video feeds” application. Our frequency minimization results show more than 40% reduction in DDoS attack success rate in the best cases when compared to the traditional periodic migration schemes on homogeneous cloud resources. The results also show that our market-driven migration location selection strategy decreases CSP cost and increases resource utilization by 30%.

Geospatial video analytics involves “data collection” of massive image/video data from the Internet of Things (IoT), and “data processing” through seamless computing at the infrastructure edge and core/public cloud platforms. Particularly in cases of (man‐made or natural) disaster incident response coordination scenarios, geospatial video analytics is valuable to provide situational awareness. The edge computing or fog computing needs to be designed with resilient performance while interfacing with the core cloud in order to cater to user's (i.e. incident commander's, first responder's) quality of experience (QoE) expectations. An additional barrier for geospatial video analytics is the fact that the distributed locations generating imagery or video content are rarely equipped with high‐performance computation capabilities to run computer vision algorithms. In this chapter, we explain the main architectural concepts of fog computing that can help in overcoming the resource scale/diversity limitations for real‐time geospatial video analytics, while also ensuring that reliability and management challenges are met.More specifically, we describe a novel cloud‐fog computing paradigm that integrates computer vision, edge routing, and computer/network virtualization areas in geospatial video analytics. We detail the state‐of‐the‐art techniques and illustrate our new/improved solution approaches based on “function‐centric” computing for the two problems of (1) high‐throughput data collection from IoT devices at the wireless edge, and (2) seamless data processing at the infrastructure edge and core cloud platforms. To assist the high‐throughput IoT data collection at the wireless edge, we present a novel deep learning‐augmented geographic edge routing that relies on physical area knowledge obtained from satellite imagery. To assist the seamless data processing, we describe a novel cloud‐fog computing framework that utilizes microservice decomposition and service chaining supported by a constrained‐shortest‐path algorithm. We conclude with a list of open challenges for adopting fog computing in geospatial video analytics for a variety of applications (e.g. face recognition in crowds, object tracking in aerial wide‐area motion imagery, reconnaissance, and video surveillance) relevant to delivering disaster‐incident situational awareness.

Patients are becoming aware of the importance of taking secure control and managing access over their medical data, thereby leading to the rise in the adoption of personal health record (PHR) systems. However, today’s PHR systems fall short in providing secure and trustable data sharing and access facilities to patients when they are in emergency situations or temporarily incapacitated. Also, the existing PHR systems are centralized and vulnerable to the single point of failure problem. Integrating PHR systems with blockchain technology can help to overcome such limitations. In this paper, we propose a blockchain-based PHR architecture that employs smart contracts to implement multi-party authorization (MPA) and threshold cryptographic schemes to automate secure and trustable medical data sharing and access in PHR systems. Moreover, we mitigate the limited storage and computation capabilities of blockchain by using InterPlanetary File System (IPFS) storage and reputation-governed trusted oracles into the proposed architecture. MPA and threshold cryptographic schemes allow the patient to split and share a secret key with a set of trusted parties, such as the healthcare regulatory agency, guardians, and hospitals, in such a way that they can collectively decide on sharing medical data on behalf of patients. We present algorithms along with their full smart contract function implementation details. We evaluate the robustness and performance of our solution by performing correctness verification and cost analysis. Furthermore, we evaluate the proposed approach in terms of security, generalization, and limitation aspects to find out its feasibility and practicality. We make our smart contract code publicly available on GitHub.

Delivering curriculum using desktop-based virtual learning environment (VLE) technologies in a collaborative group setting has been shown to reduce the social skill limitations of students with learning disabilities. However, the lack of the immersiveness and effective generalization of acquiring knowledge and skills among students remains a critical challenge in the interactive tools used in current VLEs. In this paper, we present a networked social virtual reality learning environment (VRLE) system viz., vSocial that has been redesigned based on iterative user feedback and developed in order to leverage the latest advances in integration of smart devices such as VR headsets for virtual content delivery. We describe a comparative study to evaluate technology trade-offs in the development process of transitioning from a VLE to a VRLE, from both technological and user (e.g., student/instructor) perspectives. Lastly, we outline open issues in using VRLEs which include: system complexity, emotion recognition, cybersickness and system sustainability.

Active defense schemes are becoming critical to secure cloud-based applications in the fields such as healthcare, entertainment, and manufacturing. Active defense mechanisms in cloud platforms need to be robust against targeted attacks (such as Distributed Denial-of-Service (DDoS), malware, and SQL injection) that make servers unresponsive and/or cause data breaches/loss, which in turn can cause high impact especially for healthcare applications. In this paper, we present a novel network-based active defense mechanism viz., “defense by pretense” that uses real-time attack detection and creates cyber deception e.g., by redirecting attacker’s traffic to quarantine machines and sending spoofed responses to attacker for cloud-based healthcare data processing applications. We implement our active defense mechanism by creating a realistic testbed on AWS cloud platform featuring the Observational Health Data Sciences and Informatics (OHDSI) framework for protected health data analytics with electronic health record data (SynPUF) and COVID-19 publications (CORD-19). Our evaluation experiments show the need and effectiveness of our active defense mechanism against targeted resource and data exfiltration attacks. We compare our active defense system against state-of-the-art active defense works, and our results show that our system is cost-effective, scalable and easy to deploy for active defense.

Emerging interdisciplinary data-intensive science gateway applications in engineering fields (e.g., bioinformatics, cybermanufacturing) demand the use of high-performance computing resources. However, to mitigate operational costs and management efforts for these science gateway applications, there is a need to effectively deploy them on federated heterogeneous resources managed by external Cloud Service Providers (CSPs). In this paper, we present a novel methodology to deliver fast, automatic and flexible resource provisioning services for such application-owners with limited expertise in composing and deploying suitable cloud architectures. Our methodology features a Component Abstraction Model to implement intelligent resource ‘abstractions’ coupled with ‘reusable’ hardware and software configuration in the form of “custom templates” to simplify heterogeneous resource management efforts. It also features a novel middleware that provides services via a set of recommendation schemes for a context-aware requirement-collection questionnaire. Recommendations match the requirements to available resources and thus assist novice and expert users to make relevant configuration selections with CSP collaboration. To evaluate our middleware, we study the impact of user preferences in requirement collection, jobs execution and resource adaptation for a real-world manufacturing application on Amazon Web Services and the GENI cloud platforms. Our experiment results show that our scheme improves the resource recommendation accuracy in the manufacturing science gateway application by up to 21% compared to the existing schemes. We also show the impact of custom templates knowledgebase maturity at the CSP side for handling novice and expert user preferences in terms of the resource recommendation accuracy.

Securing multi-domain network performance monitoring (NPM) systems that are being widely deployed as ‘Measurement Infrastructure-as-a-Service’ (MIaaS) in high-performance computing is becoming increasingly critical. It presents an emerging set of research challenges in cloud security given that security mechanisms such as policy-driven access to federated NPM services across multiple domains need to be designed carefully to protect MIaaS resources and data. In this paper, we advocate the design of a security middleground between default open/closed access settings and present policy-driven access controls of measurement functions for a multi-domain federation using a MIaaS. Our approach involves an analytical investigation based on a set of custom metrics to compare and contrast the legacy, role-based and more fine-grained, attribute-based access control schemes to design a security middleground. We implement the chosen middleground with a secured middleware, viz., “OnTimeSecure”. Our middleware enables ‘user-to-service’ and ‘service-to-service’ authentication, and enforces federated authorization entitlement policies for timely orchestration of MIaaS services. Lastly, we evaluate OnTimeSecure in a real multi-domain MIaaS testbed by performing threat modeling and security risk assessments to validate the analysis outcomes and demonstrate its effectiveness for easy integration and sustainable adoption.

Cloud-hosted services are being increasingly used in online businesses in e.g., retail, healthcare, manufacturing, entertainment due to benefits such as scalability and reliability. These benefits are fueled by innovations in orchestration of cloud platforms that make them programmable as Software Defined everything Infrastructures (SDxI). At the same time, sophisticated targeted attacks such as Distributed Denial-of-Service (DDoS) and Advanced Persistent Threats (APTs) are growing on an unprecedented scale threatening the availability of online businesses. In this paper, we present a novel defense system called Dolus to mitigate the impact of targeted attacks launched against high-value services hosted in SDxI-based cloud platforms. Our Dolus system is able to initiate a ‘pretense’ in a scalable and collaborative manner to deter the attacker based on threat intelligence obtained from attack feature analysis. Using foundations from pretense theory in child play, Dolus takes advantage of elastic capacity provisioning via ‘quarantine virtual machines’ and SDxI policy co-ordination across multiple network domains to deceive the attacker by creating a false sense of success. We evaluate the efficacy of Dolus using a GENI Cloud testbed and demonstrate its real-time capabilities to: (a) detect DDoS and APT attacks and redirect attack traffic to quarantine resources to engage the attacker under pretense, (b) coordinate SDxI policies to possibly block attacks closer to the attack source(s).

Applications that cater to the needs of disaster incident response generate large amount of data and demand large computational resource access. Such datasets are usually collected in real-time at the incident scenes using different Internet of Things (IoT) devices. Hierarchical clouds, i.e., core and edge clouds, can help these applications’ real-time data orchestration challenges as well as with their IoT operations scalability, reliability and stability by overcoming infrastructure limitations at the ad-hoc wireless network edge. Routing is a crucial infrastructure management orchestration mechanism for such systems. Current geographic routing or greedy forwarding approaches designed for early wireless ad-hoc networks lack efficient solutions for disaster incident-supporting applications, given the high-speed and low-latency data delivery that edge cloud gateways impose. In this paper, we present a novel Artificial Intelligent (AI)-augmented geographic routing approach, that relies on an area knowledge obtained from the satellite imagery (available at the edge cloud) by applying deep learning. In particular, we propose a stateless greedy forwarding that uses such an environment learning to proactively avoid the local minimum problem by diverting traffic with an algorithm that emulates electrostatic repulsive forces. In our theoretical analysis, we show that our Greedy Forwarding achieves in the worst case a 3 . 291 path stretch approximation bound with respect to the shortest path, without assuming presence of symmetrical links or unit disk graphs. We evaluate our approach with both numerical and event-driven simulations, and we establish the practicality of our approach in a real incident-supporting hierarchical cloud deployment to demonstrate improvement of application level throughput due to a reduced path stretch under severe node failures and high mobility challenges of disaster response scenarios.

Multi-cloud resource brokering is becoming a critical requirement for applications that require high scale, diversity, and resilience. Applications demand timely selection of distributed data storage and computation platforms that span local private cloud resources as well as resources from multiple cloud service providers (CSPs). The distinct capabilities and policies, as well as performance/cost of the cloud services, are amongst the prime factors for CSP selection. However, application owners who need suitable cyber resources in community/public clouds, often have preliminary knowledge and preferences of certain CSPs. They also lack expert guidance to handle the problem of overwhelming resource choice from CSPs, and optimization to compensate for service dynamics. In this paper, we address this challenge of optimal resource selection while also leveraging limited user’s expertise and preferences towards CSPs through multi-level fuzzy logic modeling based on convoluted factors of performance, agility, cost, and security. We evaluate the efficiency of our fuzzy-engineered resource brokering in improving allocation of resources as well as user satisfiability by using case studies and independent validations of CSPs evaluation.

With the increased push to promote data-driven methods in modern healthcare, there is a tremendous need for fast access to clinical datasets in order to pursue medical breakthroughs in the areas of personalized medicine and big data knowledge discovery. However, the inherent lack of trust between the data custodians and data consumers/users has resulted in a fully manual honest broker approach to access and process protected healthcare data. Such a manual approach leads to slow data handling, and adds to overheads needed to address data auditability and assurance needed for compliance with healthcare data security standards. In this paper, we address these challenges by proposing a trust model to enable semi-automation of the honest broker process to increase its efficiency. The trust model is based on multi-dimensional risk management principles and considers risk associated with data identifiers, as well as requestor profile and reputation. We implement and evaluate a semi-automated honest broker that uses our trust model in a community cloud testbed using the SynPUF synthetic dataset. Our experiment results show that our multidimensional risk management approach consistently identifies the lower confidentiality risk configuration in the semi-automation in comparison with a one-dimensional strategy. Thus, our semiautomated honest brokering approach improves efficiency for data custodians and data consumers by facilitation of fast and secure data access, while also ensuring compliance in the processing of the protected datasets.

Big data computation tools are vital for researchers and educators from various domains such as plant science, animal science, biomedical science and others. With the growing computational complexity of biology big data, advanced analytic systems, known as prescriptive systems, are being built using machine learning models to intelligently predict optimum computation solutions for users for better data analysis. However, lack of user-friendly prescriptive systems poses a critical roadblock to facilitating informed decision-making by users. In this paper, we detail a formative usability study to address the complexities faced by users while using prescriptive systems. Our usability research approach considers bioinformatics workflows and community cloud resources in the KBCommons framework’s science gateway. The results show that recommendations from usability studies performed in iterations during the development of prescriptive systems can improve user experience, user satisfaction and help novice as well as expert users to make decisions in a well-informed manner.

Unmanned Aerial Vehicle (UAV) systems with high-resolution video cameras are used for many operations such as aerial imaging, search and rescue, and precision agriculture. Multi-drone systems operating in Flying Ad Hoc Networks (FANETS) are inherently insecure and require efficient security schemes to defend against cyber-attacks such as e.g., Man-in-the-middle, Replay and Denial of Service attacks. In this paper, we propose a cloud-based, end-to-end security framework viz., "DroneNet-Sec" that provides secure network-edge connectivity, and computation security for drone video analytics to defend against common attack vectors in UAV systems. The DroneNet-Sec features a dynamic security scheme that uses machine learning to detect anomaly events and adopts countermeasures for computation security of containerized video analytics tasks. The security scheme comprises of a custom secure packet designed with MAVLink protocol for ensuring data privacy and integrity, without high degradation of the performance in a real-time FANET deployment. We evaluate DroneNet-Sec in a hybrid testbed that synergies simulation and emulation via an open-source network simulator (NS-3) and a research platform for mobile wireless networks (POWDER). Our performance evaluation experiments in our holistic hybrid-testbed show that DroneNet-Sec successfully detects learned anomaly events and effectively protects containerized tasks execution as well as communication in drones video analytics in a light-weight manner.

Deep Learning models such as Convolutional Neural Nets (CNNs) and Recurrent Neural Nets (RNNs) are routinely used due to their utility and benefits in various practical applications such as e.g., natural language processing and image recognition. However, despite such adoption, these techniques typically present opaqueness of their internal workings to users. The current black box approach to deep learning makes models difficult to understand and fine-tune, and the related lack of information directly influences productivity and hinders innovations in reliable and consistent model development. In this paper, we present a novel web-based, graphical approach to visualizing CNNs and RNNs to address the above adoption challenges. Our approach features an interactive graphical user interface, where the user can view the overarching network architecture and data flow, the weights and corresponding input processing at each layer, and some interpretable aspects of the model as a whole. We show the effectiveness of our visualization techniques on the MNIST dataset corresponding to an image recognition application. Our work contributes to the effective graphical visualization approaches for complex neural networks and thus makes it easier to manage, manipulate, and increase the performance of these networks.

Cloud-hosted applications are prone to targeted attacks such as DDoS, advanced persistent threats, cryptojacking which threaten service availability. Recently, methods for threat information sharing and defense require co-operation and trust between multiple domains/entities. There is a need for mechanisms that establish distributed trust to allow for such a collective defense. In this paper, we present a novel threat intelligence sharing and defense system, namely “DefenseChain”, to allow organizations to have incentive-based and trustworthy co-operation to mitigate the impact of cyber attacks. Our solution approach features a consortium Blockchain platform to obtain threat data and select suitable peers to help with attack detection and mitigation. We propose an economic model for creation and sustenance of the consortium with peers through a reputation estimation scheme that uses ‘Quality of Detection’ and ‘Quality of Mitigation’ metrics. Our evaluation experiments with DefenseChain implementation are performed on an Open Cloud testbed with Hyperledger Composer and in a simulation environment. Our results show that the DefenseChain system overall performs better than state-of-the-art decision making schemes in choosing the most appropriate detector and mitigator peers. In addition, we show that our DefenseChain achieves better performance trade-offs in terms of metrics such as detection time, mitigation time and attack reoccurence rate. Lastly, our validation results demonstrate that our DefenseChain can effectively identify rational/irrational service providers.

Bold scientific research tasks today need multi-disciplinary knowledge and interdisciplinary collaborations that require finding scholars from a particular domain with relevant knowledge. Given the variety of scholars and diversity of research tasks, finding the appropriate scholar is a critically important and challenging problem for scientific communities. Prior approaches to identify scholars use supervised learning with fixed or high-level research interest tags. Such approaches make it hard to recognize scholars with specific interests, or track their changes in research interests. Hence, there is a need to investigate suitable methods to quantify scholars' expertise knowledge for matching research tasks. In this paper, we propose a novel model viz., "ScholarFinder" that uses contextual information (abstracts or publications) for embedding a scholar's knowledge in an unsupervised learning manner. Subsequently, with pre-trained knowledge embeddings, we can perform machine learning tasks such as classification, visualization or checking whether a scholar is suitable for performing particular research tasks or not. Based on our pre-trained techniques, we also provide a novel negative sampling method to overcome the issues of missing negative samples. Using a "follow-the-money" strategy, we apply our model to a large collection of NSF (National Science Foundation) grant awards dataset collected over the last twenty years that contains more than 20,000 award records (with project abstracts and names), corresponding to 15,074 scholars who received grants. We evaluate different deep learning models to see how to use pre-trained knowledge embedding for achieving optimal performance, and how our negative sampling method improves model performance. We also compare our model with state-of-the-art baseline models (e.g., XGBoost, DNN), and our results show that the ScholarFinder model outperforms those models in terms of precision, recall, F1-score, and accuracy.

The continuous evolution of commercial Unmanned Aerial Systems (UAS) is fuelling a rapid advancement in the fields of network edge-communication applications for smart agriculture, smart traffic management, and border security. A common problem in UAS (a.k.a. drone systems) research and development is the cost related to deploying and running realistic testbeds. Due to the constraints in safe operation, handling limited energy resources, and government regulation restrictions, UAS testbed building is time-consuming and not easily configurable for high-scale experiments. In addition, experimenters have a hard time creating repeatable and reproducible experiments to test major hypotheses. In this paper, we present a design for performing tracebased NS-3 simulations that can be helpful for realistic UAS simulation experiments. We run experiments with real-world UAS traces including various mobility models, geospatial link information and video analytics measurements. Our experiments assume a hierarchical UAS platform with low-cost/high-cost drones co-operating using a geo-location service in order to provide a ‘common operating picture’ for decision makers. We implement a synergized drone and network simulator that features three main modules: (i) learning-based optimal scheme selection module, (ii) application environment monitoring module, and (iii) trace-based simulation and visualization module. Simulations generated from our implementation have the ability to integrate di↵erent drone configurations, wireless communication links (air-to-air; air-to-ground), as well as mobility routing protocols. Our approach is beneficial to evaluate network-edge orchestration algorithms pertaining to e.g., management of energy consumption, video analytics performance, and networking protocols configuration.

Industry 4.0 is the latest industrial revolution primarily merging automation with advanced manufacturing to reduce direct human effort and resources. Predictive maintenance (PdM) is an industry 4.0 solution, which facilitates predicting faults in a component or a system powered by state-of-the- art machine learning (ML) algorithms (especially deep learning algorithms) and the Internet-of-Things (IoT) sensors. However, IoT sensors and deep learning (DL) algorithms, both are known for their vulnerabilities to cyber-attacks. In the context of PdM systems, such attacks can have catastrophic consequences as they are hard to detect due to the nature of the attack. To date, the majority of the published literature focuses on the accuracy of DL enabled PdM systems and often ignores the effect of such attacks. In this paper, we demonstrate the effect of IoT sensor attacks (in the form of false data injection attack) on a PdM system. At first, we use three state-of-the-art DL algorithms, specifically, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network (CNN) for predicting the Remaining Useful Life (RUL) of a turbofan engine using NASA's C-MAPSS dataset. The obtained results show that the GRU-based PdM model outperforms some of the recent literature on RUL prediction using the C-MAPSS dataset. Afterward, we model and apply two different types of false data injection attacks (FDIA), specifically, continuous and interim FDIAs on turbofan engine sensor data and evaluate their impact on CNN, LSTM, and GRU-based PdM systems. The obtained results demonstrate that FDI attacks on even a few IoT sensors can strongly defect the RUL prediction in all cases. However, the GRU-based PdM model performs better in terms of accuracy and resiliency to FDIA. Lastly, we perform a study on the GRU-based PdM model using four different GRU networks with different sequence lengths. Our experiments reveal an interesting relationship between the accuracy, resiliency and sequence length for the GRU-based PdM models.

Multi-Unmanned Aerial Vehicle (UAV) systems with high-resolution cameras have been found to be useful for operations such as disaster management and smart farming. These systems feature Flying Ad-Hoc Networks (FANETs) that connect the computation edge with UAVs and a Ground Control Station (GCS) through air-to-ground network links. Leveraging the edge computation resources effectively with energy-awareness, and dealing with intermittent failures of FANET links are the major challenges in supporting video processing applications. In this paper, we propose a novel energy-aware dynamic computation offloading scheme for UAV systems, which provides the ability to intelligently share tasks among individual UAVs and allows for parallel execution of tasks while evenly distributing energy consumption. Intelligence gathering is performed using machine learning to create resource consumption profiles for a given set of video processing tasks prior to scheduling. Our scheme handles the problem of computation offloading tasks as a job-shop scheduling problem where we aim to minimize the total energy consumption in the edge resources while minimizing video processing times to meet application requirements. Our experimental results show our energy-aware dynamic offloading scheme enables lower processing time for low drone-to-ground server ratios and consumes less energy when compared to other offloading schemes. Notably, these results also hold in various other multi-UAV scenarios involving largely different number of detected objects.

The parameter space of CNT forest synthesis is vast and multidimensional, making experimental and/or numerical exploration of the synthesis prohibitive. We propose a more practical approach to explore the synthesis-process relationships of CNT forests using machine learning (ML) algorithms to infer the underlying complex physical processes. Currently, no such ML model linking CNT forest morphology to synthesis parameters has been demonstrated. In the current work, we use a physics-based numerical model to generate CNT forest morphology images with known synthesis parameters to train such a ML algorithm. The CNT forest synthesis variables of CNT diameter and CNT number densities are varied to generate a total of 12 distinct CNT forest classes. Images of the resultant CNT forests at different time steps during the growth and self-assembly process are then used as the training dataset. Based on the CNT forest structural morphology, multiple single and combined histogram-based texture descriptors are used as features to build a random forest (RF) classifier to predict class labels based on correlation of CNT forest physical attributes with the growth parameters. The machine learning model achieved an accuracy of up to 83.5% on predicting the synthesis conditions of CNT number density and diameter. These results are the first step towards rapidly characterizing CNT forest attributes using machine learning. Identifying the relevant process-structure interactions for the CNT forests using physics-based simulations and machine learning could rapidly advance the design, development, and adoption of CNT forest applications with varied morphologies and properties.

Unmanned Aerial Vehicles (UAVs) or drones equipped with cameras are extensively used in different scenarios such as surveillance of hazardous locations, disaster response and crime fighting. The related video streaming/analytics requires real-time drone-to-Ground Control Station (GCS) communication and computation co-ordination for desired user Quality of Experience (QoE). In situations where the quality of the video can be affected by occlusions (e.g., image distortion, frame stalling) due to network bottlenecks, there is a need to dynamically make decisions on the computation offloading and networking protocols in order to properly handle the video data for real world application purposes. In this paper, we propose a novel function-centric computing approach that helps a user to perform drone video analytics to assess a wide-area scene to chart a plan of action. Our approach involves handling network impairments affecting the switching between high resolution/low resolution video capture, or change of camera direction for assessment of the scene effectively. It also features a novel video quality enhancing algorithm based on occlusion-detection that adapts to video impairments related to image distortion and frame stalling. Our experiment results from a realistic testbed show that our approach can efficiently choose the suitable networking protocols (i.e., TCP/HTTP, UDP/RTP, QUIC) and orchestrate both the camera control on the drone, and the computation off-loading of the video analytics over limited edge computing resources. The performance improvements for computation off-loading involving our video quality enhancing algorithm are shown for different network conditions in terms of occlusion rate and processing times.

Data-intensive science applications often use federated multi-cloud infrastructures to support their compute-intensive processing needs. However, lack of knowledge about: a) individual domain's security policies, b) how that translates to application security assurance, and c) nature of performance and security trade-offs - can cause performance-security conflicts for applications and inefficient resource usage. In this paper, we propose a security-aware resource brokering middleware framework to allocate application resources by satisfying their performance and security requirements. The proposed middleware implements MCPS (Multi-Cloud Performance and Security) Broker that uses a common data model to represent applications' performance and security requirements. It performs a security-aware global scheduling to choose the optimal cloud domain, and a local scheduling to choose the optimal server within the chosen cloud domain. Using real SoyKB application workflows, we implement the proposed MCPS Broker in the GENI Cloud and demonstrate its utility through a NIST-guided risk assessment.

Cyber foraging has been shown to be especially effective for augmenting low-power Internet-of-Thing (IoT) devices by offloading video processing tasks to nearby edge/cloud computing servers. Factors such as dynamic network conditions, concurrent user access, and limited resource availability, cause offloading decisions that negatively impact overall processing throughput and end-user delays. Moreover, edge/cloud platforms currently offer both Virtual Machine (VM) and serverless computing pricing models, but many existing edge offloading approaches only investigate single VM-based offloading performance. In this paper, we propose a predictive (NP-complete) scheduling-based offloading framework and a heuristic-based counterpart that use machine learning to dynamically decide what combinations of functions or single VM needs to be deployed so that tasks can be efficiently scheduled. We collected over 10,000 network and device traces in a series of realistic experiments relating to a protest crowds incident management application. We then evaluated the practicality of our predictive cyber foraging approach using trace-driven simulations for up to 1000 devices. Our results indicate that predicting single VM offloading costs: (a) leads to near-optimal scheduling in 70% of the cases for service function chaining, and (b) offers a 40% gain in performance over traditional baseline estimation techniques that rely on simple statistics for estimations in the case of single VM-offloading. Considering a series of visual computing offloading scenarios, we also validate our approach benefits of using online versus offline machine learning models for predicting offloading delays.

IoT (Internet of Things) based smart devices such as sensors have been actively used in edge clouds i.e., ‘fogs’ along with public clouds. They provide critical data during scenarios ranging from e.g., disaster response to in-home healthcare. However, for these devices to work effectively, end-to-end security schemes for the device communication protocols have to be flexible and should depend upon the application requirements as well as the resource constraints at the network-edge. In this paper, we present the design and implementation of a flexible IoT security middleware for end-to-end cloud–fog communications involving smart devices and cloud-hosted applications. The novel features of our middleware are in its ability to cope with intermittent network connectivity as well as device constraints in terms of computational power, memory, energy, and network bandwidth. To provide security during intermittent network conditions, we use a ‘Session Resumption’ algorithm in order for our middleware to reuse encrypted sessions from the recent past, if a recently disconnected device wants to resume a prior connection that was interrupted. In addition, we describe an ‘Optimal Scheme Decider’ algorithm that enables our middleware to select the best possible end-to-end security scheme option that matches with a given set of device constraints. Experiment results show how our middleware implementation also provides fast and resource-aware security by leveraging static properties i.e., static pre-shared keys (PSKs) for a variety of IoT-based application requirements that have trade-offs in higher security or faster data transfer rates.

New paradigms such as Mobile Edge Computing (MEC) are becoming feasible for use in, e.g., real-time decision-making during disaster incident response to handle the data deluge occurring in the network edge. However, MEC deployments today lack flexible IoT device data handling such as handling user preferences for real-time versus energy-efficient processing. Moreover, MEC can also benefit from a policy-based edge routing to handle sustained performance levels with efficient energy consumption. In this paper, we study the potential of MEC to address application issues related to energy management on constrained IoT devices with limited power sources, while also providing low-latency processing of visual data being generated at high resolutions. Using a facial recognition application that is important in disaster incident response scenarios, we propose a novel “offload decision-making” algorithm that analyzes the tradeoffs in computing policies to offload visual data processing (i.e., to an edge cloud or a core cloud) at low-to-high workloads. This algorithm also analyzes the impact on energy consumption in the decision-making under different visual data consumption requirements (i.e., users with thick clients or thin clients). To address the processing-throughput versus energy-efficiency tradeoffs, we propose a “Sustainable Policy-based Intelligence-Driven Edge Routing” algorithm that uses machine learning within Mobile Ad hoc Networks. This algorithm is energy aware and improves the geographic routing baseline performance (i.e., minimizes impact of local minima) for throughput performance sustainability, while also enabling flexible policy specification. We evaluate our proposed algorithms by conducting experiments on a realistic edge and core cloud testbed in the GENI Cloud infrastructure, and recreate disaster scenes of tornado damages within simulations. Our empirical results show how MEC can provide flexibility to users who desire energy conservation over low latency or vice versa in the visual data processing with a facial recognition application. In addition, our simulation results show that our routing approach outperforms existing solutions under diverse user preferences, node mobility, and severe node failure conditions.

The Applications and Services in the Year 2021 workshop was successfully organized on January 27--28, 2016 in Washington DC through funding support from the National Science Foundation (NSF). The goal of the workshop was to foster discussions that bring together applications researchers in multidisciplinary areas, and developers/operators of research infrastructures at both national, regional, university and city levels. Discussions were organized to identify grand challenge applications and obtain the community voice and consensus on the key issues relating to applications and services that might be delivered by advanced infrastructures in the decade beginning in 2020. The timing and organization for the workshop is significant because today's digital infrastructure is undergoing deep technological changes and new paradigms are rapidly taking shape in both the core and edge domains that pose fundamental challenges. The key outcomes of the discussions were targeted to enhance the quality of peoples' lives while addressing important national priorities, leveraging today's cutting edge applications such as the Internet of Things, Big Data Analytics, Robotics, The Industrial Internet, and Immersive Virtual/Augmented Reality. This report summarizes the workshop efforts to bring together diverse groups for delivering targeted short/long talks, sharing latest advances, and identifying gaps that exist in the community for 'research' and 'infrastructure' needs that require future NSF funding.

Campuses are increasingly adopting hybrid cloud architectures for supporting data-intensive science applications that require “on-demand” resources, which are not always available locally on-site. Policies at the campus edge for handling multiple such applications competing for remote resources can cause bottlenecks across applications. These bottlenecks can be proactively avoided with pertinent profiling, monitoring and control of application flows using software-defined networking and pertinent selection of local or remote compute resources. In this paper, we present an “application-driven overlay network-as-a-service” (ADON) that manages the hybrid cloud requirements of multiple applications in a scalable and extensible manner by allowing users to specify requirements of the application that are translated into the underlying network and compute provisioning requirements. Our solution involves scheduling transit selection, a cost optimized selection of site(s) for computation and traffic engineering at the campus-edge based upon real-time policy control that ensures prioritized application performance delivery for multi-tenant traffic profiles. We validate our ADON approach through an emulation study and through a wide-area overlay network testbed implementation across two campuses. Our workflow orchestration results show the ADON effectiveness in handling temporal behavior of multi-tenant traffic burst arrivals using profiles from a diverse set of actual data-intensive applications.

Multi-domain end-to-end network performance monitoring federations such as perfSONAR are increasingly being used in Big Data application management. They rely on trustworthy collaborative measurement intelligence to identify and diagnose network anomaly events that impact application performance. Large volumes of end-to-end measurement traces are generated on a daily basis, and new Big Data analysis techniques are needed to isolate network-wide anomaly event(s) and to diagnose the root-cause(s). In addition, not all network operators and application users have enough knowledge and experience to understand the anomaly events. The lack of a platform for sharing knowledge and working collaboratively makes it difficult to isolate and diagnose network-wide anomaly events quickly and accurately. In this paper, we define a “social plane” that relies on recommended measurements based on “content-based filtering” and “collaborative filtering” approaches to enable network performance expectation management. Based on similarity analysis, the content-based filtering facilitates users to subscribe to useful measurements, and the collaborative filtering promotes users to share knowledge on anomaly symptoms. Using real perfSONAR measurements and synthetic events, we show the effectiveness of our social plane approach within a SoyKB Big Data application case study using social network creation and mingling of experts. Our experimental results show that our measurements recommendation scheme has high precision, recall, and accuracy, as well as efficiency in terms of the time taken for large volume measurement trace analysis.

This paper shows how to properly achieve elasticity for network firewalls deployed in a cloud environment. Elasticity is the ability to adapt to workload changes by provisioning and de-provisioning resources in an autonomic manner, such that at each point in time the available resources match the current demand as closely as possible. Elasticity for cloud-based firewalls aims to satisfy an agreed-upon performance measure using only the minimal number of cloud firewall instances. Our contribution lies in determining the number of firewall instances that should be dynamically adjusted in accordance with the incoming traffic load and the targeted rules within the firewall rulebase. To do so, we develop an analytical model based on the principles of Markov chains and queueing theory. The model captures the behavior of a cloud-based firewall service comprising a load balancer and a variable number of virtual firewalls. From the analytical model, we then derive closed-form formulas to determine the minimal number of virtual firewalls required to meet the response time specified in the service level agreement. The model takes as input key system parameters including workload, processing capacity of load balancer and virtual machines, as well as the depth of the targeted firewall rules. We validate our model using discrete-event simulation, and real-world experiments conducted on Amazon Web Services cloud. We also provide numerical examples to show how our model can be used in practice by cloud performance/security engineers to achieve proper elasticity under fluctuating traffic load and variable depth of targeted firewall rules.

There is an increasing demand to develop innovations in eldercare technologies that can be delivered as 'Apps' at a cloud-scale to facilitate proactive monitoring and targeted care coordination. This article presents the design for an 'ElderCare-as-a-SmartService' (ECaaS) system that integrates Apps for in-home health monitoring, and remote physical therapy coaching, respectively. The focus is on the transformation of the Apps into a cloud-based living lab, which then enables on-going App development/refinement to realize a real-world enhanced living environment for eldercare that is secure, privacy-preserving and socially embedded. We detail the system and network requirements for cloud-based delivery of ECaaS Apps featuring sensor data analysis and interactive interfaces across urban, suburban and rural areas. We conclude with a proposal for sociotechnical investigations that can be lead to a human-centered design that engages older adults and care coordinators as co-developers of ECaaS, while also simultaneously improving their quality of life.

Scientific workflows due to their data and memory intensive requirements are among the prime applications which benefit by leveraging cloud computing. However, Cloud service providers (CSPs) have distinct policies and service dynamics that present a problem of excess choice for users. Performance and cost of the cloud services are among the principal factors in CSP selection for scientific bioinformatics workflows. The workflows typically are based on private data, and require diverse cloud resources, thus often requiring synergistic services from multiple CSPs. In this paper, we address this challenge of multi-cloud resource selection using cloud template solutions based on user specifications. We propose an optimizer that incorporates a combinatorial optimization model built on performance, cost and CSPs interoperability factors. The optimizer is integrated within a novel resource broker (i.e., OnTimeURB) for prescriptive recommendations of template solutions with intuitive choices for users. We implement and evaluate the OnTimeURB recommendations framework with a catalog of bioinformatics workflow applications integrated within a KBCommons science gateway. The evaluation considered four CSP resources featuring more than 300 different machine configuration instances. Our evaluation results show that our OnTimeURB creates consistently more economical, performance optimized and practical cloud solutions compared to a k-nearest neighbors (k-NN) approach.

Objective: Search and rescue after mass casualty incidents relies on robust data infrastructure. Federal Emergency Management Agency (FEMA’s) Task Force 1 (TF1) trains its volunteers to locate and virtually tag scene incidents using a global positioning satellite (GPS) device programmed with markers for each incident (Iron Sights). The authors performed a pilot study comparing Iron Sights™ to a Wi-Fi-based real-time incident geolocation and virtual tagging dashboard (Panacea™) in creating a dynamic common operating picture. Design: Twenty-nine stations were placed at a predefined scene incident, each featuring a set of varying waypoint markers using standard FEMA / TF1 nomenclature. Two volunteers performed the experiment for both the Iron Sights and Panacea systems, digitally tagging all station waypoints. Setting: TF1 simulation training field. Main outcome measure(s): Metrics compared included GPS location precision, marker accuracy, and delay between scene sweep and common operational picture (COP) generation. Results: Two hundred and sixty-one waypoints were digitally tagged after excluding three stations for missing data. The average GPS location difference for all waypoints between Iron Sights and Panacea was 3.65 m. Marker tagging accuracy between Iron Sights and Panacea was equivalent and not statistically different (78.8 percent vs 66.2 percent, respectively, p = 0.11). Waypoints were tagged in 26.59 minutes and 10.55 minutes on average, respectively. Time from scene sweep to virtual COP generation was 7.97 minutes for Iron Sights after complete scene sweep and 37 seconds for Panacea for each waypoint posting in real-time. Conclusions: Panacea generated the COP in real-time compared to a delay with Iron Sights while maintaining the same location precision and marker accuracy. This pilot trial successfully demonstrated the ability to provide real-time actionable intelligence to incident commanders during mass casualty search and rescue missions. Larger field trials are recommended to refine the system and broaden its capabilities.

Cloud computing has become a necessary utility for scientific and technical applications. Many diverse web services are published and subscribed using cloud data centers. It has become fairly easy to use services from Cloud Service Providers (CSPs) for computation and data processing. However, even with all their benefits, commercial cloud resources are not economical when large data processing is required. Hence, educators and researchers need guidance to use commercial cloud resources to run large data processing workflow applications within a budget. In this paper, we propose a framework to help users to leverage distributed compute resources to execute data-intensive application workflows, under budget constraints. We demonstrate how our framework can be used by users who may have access to small-scale compute resources in-house, to seamlessly interoperate with public cloud resources.

Unmanned aerial vehicles (UAV) or drone systems equipped with cameras are extensively used in different surveillance scenarios and often require real-time control and highquality video transmission. However, unstable network situations and various transport protocols may result in impairments during video streaming, which in turn negatively impacts user’s quality of experience (QoE). In this paper, we propose a dynamic computation offloading and control framework, named DyCOCo, based on image impairment detection under various available network bandwith conditions. Our DyCOCo framework demo features IoT devices in a testbed setup on the GENI infrastructure. Our demo results show that our DyCOCo approach can efficiently choose the suitable networking protocols and orchestrate both the camera control on the drone, and the computation offloading of the video analytics over limited edge computing/networking resources.

Computer vision applications are increasingly used on mobile Internet-of-Things (IoT) devices such as drones. They provide real-time support in disaster/incident response or crowd protest management scenarios by e.g., counting human/vehicles, or recognizing faces/objects. However, deployment of such applications for real-time video analytics at geo-distributed areas presents new challenges in processing intensive media-rich data to meet users’ Quality of Experience (QoE) expectations, due to limited computing power on the devices. In this paper, we present a novel policy-based decision computation offloading scheme that not only facilitates trade-offs in performance vs. cost, but also aids in offloading decision to either an Edge, Cloud or Function-Centric Computing resource architecture for real-time video analytics. To evaluate our offloading scheme, we decompose an existing computer vision pipeline for object/motion detection and object classification into a chain of container-based micro-service functions that communicate via a RESTful API. We evaluate the performance of our scheme on a realistic geo-distributed edge/core cloud testbed using different policies and computing architectures. Results show how our scheme utilizes state-of-the-art computation offloading techniques to Pareto-optimally trade-off performance (i.e., frames-per-second) vs. cost factors (using Amazon Web Services Lambda pricing) during real-time drone video analytics, and thus fosters effective environmental situational awareness.

Virtual Reality (VR)-based Learning Environments (VRLEs) are gaining popularity due to the wide availability of cloud and its edge (a.k.a. fog) technologies and high-speed networks. Thus, there is a need to investigate Internet-of-Things (IoT)-based application design concepts within social VRLEs to offer scalable, cost-efficient services that adapt to dynamic cloud/fog system conditions. In this paper, we investigate the cost-performance trade-offs for an IoT-based application that integrates large-scale sensor data from Social VRLEs and coordinates the real-time data processing and visualization across cloud/fog platforms. To facilitate dynamic performance adaptation of the IoT-based application with increased user scale, we present a set of cost-aware adaptive control rules. The implementation of the rules is based on an analytical queuing model that determines the performance states of the IoT-based application, given the current workload and the allocated cloud/fog resources. Using the IoT-based application in an exemplar VRLE use case, we evaluate the cost-performance trade-offs with three system architectures i.e., cloud-only, edge-only and edge-cloud architectures. Experiment results illustrate the best/worst practices in the cost-performance trade-offs for a range of simulated IoT scenarios involving monitoring user emotional data collected by using brain sensors. Our results also detail the impact of the system architecture selection, and the benefits in enabling feedback about student emotions to instructors during Social VR learning sessions. Lastly, we show the benefits of integrating our model-based feedback control in maximizing IoT-based application performance while keeping the associated costs at a minimum level.

Traditionally, Network Function Virtualization uses Service Function Chaining (SFC) to place service functions and chain them with corresponding flows allocation. With the advent of Edge computing and IoT, a retiable composition of latency-sensitive SFCs is needed to support applications in geo-distributed cloud infrastructures. However, the optimal SFC composition in this case becomes the NP-hard integer multi-commodity-chain flow (MCCF) problem that has no known approximation guarantees. In this paper, we present a novel practical and near optimal SFC composition approach for geo-distributed cloud infrastructures that also admits end-to-end network QoS constraints such as latency, packet loss, etc. Specifically, we propose a novel metapath composite variable approach that reaches 99% optimality on average and takes seconds for practically sized integer MCCF problems of US Tier-1 ($\sim$300 nodes) and regional ($\sim$600 nodes) infrastructure providers’ topologies. To ensure reliability, we compose SFCs with capacity chance-constraints and backup policies. Using trace-driven simulations comprising of challenging disaster-incident conditions, we show that our solution composes twice as many SFCs than the state-of-the-art network virtualization methods.

Communication and coordination in a mass casualty disaster scenario is limited and difficult for medical personnel in the absence of necessary communication infrastructure and technologies that enable situation awareness. This leads to misdirected and delayed triage of scene-wide critically injured patients, especially when there are large numbers of patients needing diverse care levels. A hierarchical cloud-fog setup involving a core cloud and edge clouds (i.e., fogs) can help the patient triage related real-time data movement challenges. It can create visual situational awareness and overcome infrastructure limitations at the wireless network edge. In this paper, we describe our “Panacea's Cloud”, a hierarchical cloud-fog platform that provides augmented reality benefits with real-time human communication and geolocation services through the integration of a standardized Incident Command System (ICS) with smart devices such as heads-up displays, virtual beacons, and wireless mesh network elements. We show how the ICS can run using a core cloud as an intelligent dashboard that leverages fog resources and data collection close to the incident scenes. In addition, we describe experiment results from evaluations of Panacea's Cloud dashboard usability and platform performance that show benefits and issues in hierarchical cloud-fog service management in support of ICS communication during disaster incident coordination.