Download Research Tools
In my previous blogs, I talked about the Lab of Things (LoT), which is a flexible platform for conducting experimental research using connected devices in homes and beyond. Since LoT’s beta release in July 2013, we have made a lot of progress on two fronts. First, we’ve been busily extending and perfecting various features of LoT; second, we’ve been working closely with academic partners to help them adopt and deploy LoT for their research. Right from the inception of the project, we have been working closely with our academic collaborators to understand better the needs of such a platform. One of the key requirements was to be able to support both off-the-shelf and custom devices. Today, LoT supports a host of off-the-shelf devices, including Z-Wave switches and multisensors, the Philips Hue light bridge, the Kinect for Windows sensor, and the Envi energy sensor. The driver model is extensible and adding a new device involves learning design patterns from the existing drivers (which are released in source form) and writing new ones. For developing custom devices and sensors, LoT now supports robust drivers for .NET Gadgeteer.We have made significant progress toward making HomeOS (the client-side component of LoT) more robust and extensible. Meanwhile, on the cloud-services side, LoT now includes a management portal that allows near real-time monitoring of the status of home hubs and enables researchers to monitor deployments separately for each study. The user interface of applications running in each home hub are now available securely from anywhere by using any device with a modern web browser.
As part of the Lab of Things, we have a system for the efficient storage and sharing of data across applications running on LoT. This storage system offers the abstraction of a stream of time-tag-value records, with arbitrary, application-defined tags, and it supports efficient querying based on time or tags. The Lab of Things file system uses cloud storage as a seamless extension of local storage. It builds an index on the data stream and organizes the data into chunks of multiple records, which enables efficient compression, encryption, storage, and transfer of data.
In the following video, Ratul Mahajan, a Microsoft researcher working on the project, talks about the motivation of the project and current capabilities. Subsequently, AJ Brush, another Microsoft researcher working on this project, gives a demo of the LoT client-side set up using various devices.
In addition, we have been working with a number of academic researchers and students to enable them to adopt the Lab of Things and use it to deploy their experiments. A case in point is the ongoing work at University College London, where a dedicated team of students developed an analytics engine for the Lab of Things. The source code of the analytics engine is available on CodePlex to use and extend. Professor Affan Syed and his students at FAST-NUCES, Pakistan, are using the Lab of Things to develop a system to optimize and control the use of electrical power in homes to help address the acute shortage of energy in that country. The team is busy scaling up the project to deploy in a large number of homes with the goal to understand energy usage and optimization goals in a wider cross section of society. This video demonstrates the system.
Not only for research projects, the Lab of Things is also being used for teaching. Professor Nilanjan Banerjee from University of Maryland at Baltimore County is offering a Lab of Things-based graduate-level course on Systems for Smart Home Automation, in which students will study the challenges in smart home automation systems and use the Lab of Things to build software systems for smart home automation.
The Lab of Things website lists additional LoT projects that the academic community is implementing. See the academic projects page. —Arjmand Samuel, Senior Research Program Manager, Microsoft Research ConnectionsLearn more
Top of page
Today, February 5, 2014, marked the kickoff workshop for the Swiss Joint Research Center (Swiss JRC), a collaborative research engagement between Microsoft Research and the two universities that make up the Swiss Federal Institutes of Technology: ETH Zürich (Eidgenössische Technische Hochschule Zürich, which serves German-speaking students) and EPFL (École Polytechnique Fédérale de Lausanne, which serves French-speaking students).
The Swiss JRC is a continuation of a collaborative engagement that began five years ago, when these same three partners embarked on ICES (Innovation Cluster for Embedded Software). In renewing our collaboration, we have broadened and deepened the computer science engagements, as we chart a course for another five years of research. During the two-day workshop at Microsoft Research Cambridge, we will launch seven new projects that constitute the next wave of research collaborations for the Swiss JRC. Today, we heard EPFL’s Edouard Bugnion describe the planned work of the Scale-Out NUMA project, which involves the study of the computer architectural and system software implications of aggressive scale-out, energy-efficient computing in datacenters.
Workshop speakers, listed clockwise from upper left: Daron Green, Andrew Blake, James Larus (EPFL), and Markus Püschel (ETH Zürich)
Now I’m looking forward to tomorrow’s sessions, especially the presentation by Otmar Hilliges (ETH Zürich), who will discuss the fascinating topic of human-centric flight. This proposed research seeks to create an entirely new form of interactive systems, leveraging micro-aerial vehicles (MAVs), also known as flying robots, to create novel user experiences. This project could have a profound impact on our future ability to navigate environments that are inhospitable to people or standard land-based robots.
Attendees of the kickoff workshop for the Swiss JRC
The following seven projects will be launched at the workshop:
Scale-Out NUMAEdouard Bugnion, EPFLBabak Falsafi, EPFLDushyanth Narayanan, Microsoft ResearchMicro-Aerial Vehicles (MAVs) for Interaction, Videography, and 3D ReconstructionOtmar Hilliges, ETH ZürichMarc Pollefeys, ETH ZürichShahram Izadi, Microsoft Research Software-Defined Networks: Algorithms and Mechanisms Roger Wattenhofer, ETHZRatul Mahajan, Microsoft Research
Investigation into fundamental issues concerning software-defined networks and how they can be tackled using a game theory approach
Efficient Data Processing Through Massive Parallelism and FPGA-Based AccelerationGustavo Alonso, ETH ZürichKen Eguro, Microsoft Research
Exploration of efficient implementation of FPGAs as co-processors in data centers and support for database querying
Authenticated Encryption: Security Notions, Constructions, and ApplicationsSerge Vaudenay, EPFLIlya Mironov and Markulf Kohlweiss, Microsoft Research
Developing enhanced security notions for authenticated encryption schemes and proving that they are secure
Towards Resource Efficient Data CentersFlorin Dinu, EPFLSergey Legtchenko, Microsoft Research
Researching how memory can be best utilized in homogeneous computational situations, where the operating system must handle parallel, data-intensive tasks
Availability and Reliability as a Resource for Large-Scale in Memory Databases on Datacenter ComputersTorsten Hoefler, ETHZMiguel Castro, Microsoft Research
Researching new approaches to building resilience and predicting resilience in systems with more economical, lower levels of redundancy
These projects represent some of the most interesting and engaging research challenges in Microsoft Research’s broad portfolio of university partnerships. I particularly value the opportunity to share our domain expertise in these open collaborations with two of the world’s top computer-science research departments. All three organizations bring unique perspectives and great talent to the collaboration, and all focus on solving tough technical challenges in areas as diverse as human-computer interaction, machine vision, performance and energy scalability, mobile computing, and data center optimization.I’ll keep you up to date on this journey over the coming months and years, as the Swiss JRC works to accelerate scientific discoveries and breakthroughs that push the boundaries of our imagination.—Daron Green, Senior Director, Microsoft Research ConnectionsLearn more
As regular readers of this blog know, the Windows Azure for Research program recurrently solicits proposals on the use of Windows Azure, Microsoft’s cloud-computing platform, in scholarly research. Winning projects receive a one-year allocation of Windows Azure storage and compute resources.
We review these proposals on the fifteenth of even-numbered months (February, April, June, and so forth), so the next deadline, February 15, is fast approaching. This marks our third round of solicitations, and the response so far has been outstanding, as a review of current grantees and their projects attests.
In addition to these standing, bi-monthly requests for proposals, we are initiating a new set of calls, focused on specific cloud-based research topics. Submissions for the first of these special calls are due on April 15, 2014.
Our first special call—Science VMs for Research—requests proposals to build virtual machine (VM) images that can be shared with communities of users. While it is standard practice for scientific communities to share important open-source, domain-specific software tools, using these tools often involves complex installation procedures or the resolution of library conflicts. Cloud computing obviates such impediments by enabling communities to share a complete operating system image, pre-installed with all the tools needed by specialized groups of users. Thus, a newcomer to the group can install the image in the cloud and be doing productive work very quickly. Moreover, the community can keep the cloud-based VM image updated with the latest version of the software.
Microsoft Open Technologies operates VM Depot, a community-driven catalog of preconfigured operating systems, applications, and development stacks—VM images that can installed in minutes by anyone with a Windows Azure account. Several VM Depot images have proven popular with the scientific community. For example, Elastacloud has donated an image called Azure Data Analysis, which includes R, IPython, and a number of high quality open-source, data analysis tools. Several other domain-specific VMs are in the works.
The Science VMs for Research call will provide grants of Windows Azure resources to develop and test new contributions to the VM Depot. Submit your proposals for the special call via our submission site; proposals should include “Science VM” in the project title and must be received by April 15.
We’re looking forward to reviewing both the February 15 and April 15 proposals, as we work together to bring the power of cloud computing to scholarly and scientific research.
—Dennis Gannon, Director of Cloud Research Strategy, Microsoft Research Connections