Elastic Mobile Sensing as a Service

Mobile Sensing is one of the IoT core enabling technologies. There is currently no platform that offers mobile sensing as a a service due to the lack of interoperability and the large diversity of heterogenous sensing resources. My research builds a generic platform that offers mobile sensing as a service. The platform disassociates sensor functionality from application demands. Multipurpose sensing objects (e.g., smartphones) can be flexibly utilized in various ways according to the application's varying demands and required sensing functionality. The platform also decouples sensor worker recruiting, sensor data collection, data analytics services, and service delivery. The platform pioneers the idea of sensor data marketplace, where sensor workers, recruiters and interested users (applications or entities) meet to negotiate and perform service-oriented interactions. We device a set of service broker APIs that enable third-party value-added services to easily integrate with the platform.

QuARAM: QoS-aware cloud application management

Many business enterprises consider moving their IT infrastructure to the cloud for potential cost saving and flexible resource management. However, deploying and running applications on the cloud is still challenging and requires lots of planning to make the right decision. In addition, managing could applications after deployment is inefficient due to the lack of cross-platform interoperability and management. To make the cloud more attractive, cloud application management needs to become more provider-independent, autonomic and Quality-of-Service (QoS) aware. We propose the QuARAM framework for QoS-aware autonomic cloud application management. QuARAM supports application developers in selecting a cloud provider, provisioning resources on the provider, deploying the application, and then managing the execution of the application. We are currently focusing on two aspects: 1) how to select the cloud provider that best suits a specific application; 2) how to efficiently manage cloud applications to ensure that Service Level Agreements (SLAs) are satisfied. We developed QuARAMRecommender, a cloud service recommender framework that supports autonomic service selection. QuARAMRecommender makes selection decisions based on the application requirements, customer preferences, and provider reputation. The system implements machine learning techniques (e.g., reinforcement learning) to improve the quality of recommendations. For cloud application monitoring, we built SLAM, a customizable provider-independent monitoring platform that enables interoperability across federated clouds. It generates user-defined monitoring templates based on SLA high-level parameters and deploys monitoring agents on distributed cloud nodes.

Quality of Big Data in mobileSensing

Mobile devices are a cornerstone in the emerging crowdsourcing and participatory sensing computing paradigms. Mobile devices contribute a wide variety of information, ranging from characterizing the behavior or users to real time access to context information, capitalizing on their mobility and sensing capabilities. The size and diversity of such collected data and the implications of mobile environments pose a real challenge on data quality. This project addresses quality assurance and data management in crowdsourcing BigData. Prime challenges that are currently investigated include recruitment and allocation of data sources, data cleaning, data fusion, scamming detection, and query semantics and optimization.

Advancing the architecture of mobile service delivery

Pervasive and ubiquitous computing is the trend towards connecting people with their surrounding objects, creating smart environments. New scenarios and applications in pervasive computing emerge everyday, which offer enormous business potential to mobile Web service provisioning. This project looks at enhancing the user experience of mobile Web services through a paradigm shift for mobile devices to not only act as Web clients, but also as Web services providers. The project focuses on provisioning personalized and reliable mobile services and applications, incorporating user context, such as location, preferences, device capabilities, and environment context in order to achieve a differential user experience. The project expands in two directions: 1) Personal Services, where users are able to provide user-specific services at anytime, anywhere to a set of authorized customers with a custom level of user-controlled access regulations; 2) Cloud-Assisted mobile services, where the cloud plays a pivotal role to relieve the burden on mobile resources, while bridging the gaps between resource-constrained devices and computational-intensive applications. This allows applications to benefit from the mobility and dynamic context access that mobile devices can offer, while simultaneously taking advantage of resource-rich environments.

Mobile apps as a Service

In this project we look at how we can customize the behaviour of mobile applications based on the user's context and preferences. AppaaS is a context-aware system for provisioning mobile applications as a service, exploiting several context information including location information, user preferences and access rights, device profile, customer ratings, and current time to provision the best relevant mobile application to such a context. AppaaS supports state preservation, where application-specific data that is relevant to a user is stored for the user's future reference. A prototype is deployed on the cloud to demonstrate the system performance with respect to finding relevant applications to a specific context and controlling the application's functions according to the user's access privileges.

Securing mobile data in healthcare systems

In the healthcare domain, there is a drive to develop mobile healthcare applications to proactively collect and disseminate information. This concept of "healthcare-as-a-mobile-service" transforms a patient's smartphone into an active data provider that a healthcare giver can query to retrieve information, collected via body sensors, regarding the patient's health status. However, smartphones have storage capacity limitations making implementing secure healthcare applications challenging. In this project we study with Anne Kayem (University of Cape Town, South Africa) the possibilily of securing mobile data exchange in healthcare services. We are trying to employ data fragmentation and caching to maximize storage utilization and enforce data security. The target is to create data fragments with similar information to facilitate query processing. We then can use a hierarchical caching structure to define a prioritization mechanism that is useful in determining, based on relevance and frequency of access, which fragments of the data to outsource or store on the mobile phone.