Banner Research

Header Current Projects
the SWAN The SWAN( project aims at building a scalable and robust sensing framework for smartphones. With the help of such a framework, application developers can easily build context-based apps in the field of health care, transportation, fire and safety etc. The framework uses a formal language (called SWAN-Song) to create sensor-based expressions. It already supports more than 20 sensors and new sensors can be easily plugged in. Distributed sensing and Cloud support are two other main features of this framework.

BaTSBaTS is a user-centric cloud scheduling framework, recommending the optimal cloud resources configurations for a given application in terms of performance (makespan or throughput) and cost (monetary or energy). Using a variety of algorithms (genetic algorithms, hill climbing, bounded knapsack), BaTS computes fast and accurate approximations of the Pareto fronts of such cloud configurations.

squirrelSquirrel is a caching system for fast and scalable deployment of virtual machines in infrastructure clouds. Squirrel is built using novel techniques such as block-level caching of boot working sets, scalable deduplication, and snapshot/resume combined with resource hot-plugging.

SENSei At Sensei, we aim to improve health through behavioral change and compliance, by improving the personal motivation of people, using behavioral intervention technology based on modern ICT, especially smartphones, off-the-shelf sensors and clouds. Our project differs from existing work in several ways:

• It combines all three aspects of health: physical, mental and social, and their interactions.

• It provides real-time feedback to users while they are exercising, using a smartphone based sensor network in combination with cloud technology; we thus provide a complete sense-model-action loop.

• It provides personalized feedback based on individual goals, but also exploiting aggregated information from groups of users to enhance the feedback.

We propose to attain these goals by implementing an intelligent framework that combines information from all of the available sensors in order to coach each user based on her individual needs and physical parameters.

CloudBnB is a resource management framework which executes applications in nested VMs running within VMs allocated from cloud providers. CloudBnB reconciliates tenant objectives, in terms of application performance and reduced (paid-for) resource wastage, with cloud provider’s objective of reducing their operational costs. The key technology behind CloudBnB is nested virtualization. Nested virtualization allows tenants to implement their desired resource management policies while cloud providers can focus on efficient and scalable resource allocation. Due to beneficial consolidation opportunities in terms of application performance as well as economy of scale, CloudBnB typically favors scaling up to larger VMs to host the tenants' (nested) VMs. This further reduces cloud provider’s management overhead and allows for more cost effective physical resource allocation.