European Microsoft Innovation Center – Our Research Areas

At the European Microsoft Innovation Center (EMIC), our research is currently focused on three areas: Mobile, Home, and Enterprise. In each of these areas, our research encompasses multiple topics, as described below.

1Mobile

1.1Sensor and Context Management

Sensors are routinely used these days to collect information about the environment. This sensor and context information can be used by applications to adapt their behaviour in scenarios as diverse as healthcare, care for the elderly, cars, home automation, environmental control, agriculture, and defense systems. Imagine you enter a hotel room, and your mobile device discovers sensors in the room telling you which hotel room you are in and the services the hotel offers, like setting the room temperature and your wake-up call, making dinner reservations, or loading your favourite pictures on a wall screen.

Our Sensor and Context Management research aims to make sensor and context information an easy-to-use technology for mobile applications by making them aware of the environment they are executed in and providing information about the situation of their users.

WEAR-IT@work, ASK-IT, EMERGE, and WASP are examples of European projects where we work on sensor and/or context management.

Back to top

1.2Mobile Peer-to-Peer

Peer-to-peer (P2P) technology allows collaboration and sharing of information between multiple nodes without a central server. In a P2P network, the traditional view of clients and servers is replaced by one of peer nodes, which can effectively function as both client and server.

Implementing a P2P system from scratch is a complicated task because the system is fully distributed and needs to be able to scale and work properly regardless of errors or intermittent connectivity of nodes.

The Microsoft P2P software development kit (SDK) has been developed to solve this issue and provide a basic framework allowing programmers to write true P2P applications without having to worry about low-level networking issues. Please note the P2P SDK has been developed with desktop nodes in mind and has not considered how mobile devices could participate in P2P networks. Issues such as battery, processing power, storage, and connection limitations have not been addressed, and, for these reasons, the design of the P2P SDK is inappropriate for mobile devices

The mission of our current mobile P2P research is to study the limitations of the existing desktop implementation and provide the modifications and new features required to integrate mobile devices seamlessly, while preserving full compatibility with the existing application programming interface.

Back to top

1.3Network Awareness

Due to the rapid evolution of wireless networks and their steadily increasing dissemination, mobile devices are going through rapid changes in their support for wireless networks. Our mission is to make the radio embedded on mobile devices smarter.

Multi-radio devices are already common. Mobile devices now offer several embedded wireless network technologies such as Wi-Fi, Bluetooth, and cellular (GSM, CDMA…). In the very near future, we can expect this trend to continue with the introduction of other wireless technologies such as WiMax and UWB.

Because of increasing processor speeds, people are also talking about Software Defined Radio (SDR). An SDR uses a reconfigurable hardware platform allowing multiple radio standards to be supported. We want to explore how these new radio architectures can be integrated into operating systems and, conversely, how these new architectures will affect operating system architectures. A second objective is to simplify MAC layer development by using new software development methodologies, tools, and run-times.

Cognitive Radio represents one way forward, where RF information is combined with user needs based on contextual information to provide radio resources most appropriate for those needs. Based on current wireless networking architectures (multi radios) and future SDR-based architectures, EMIC wants to explore cognitive support for network selection and usage. Depending on the application, user needs, and policies, the most appropriate network should be selected. This process should be a self-organized process, since current network selection and configuration is far too complicated for normal users. It should include contextual knowledge of the environment, the user, and the applications. Optimizations of spectrum use and power consumption are also important topics that will have to be considered in this environment.

To fulfill our mission, the network awareness research team is working on improving network selection based on user context information, radio information, application requirements, minimal power consumption, and other policies. To achieve our goal of providing an easy user interface, we are also working on user-friendly concepts for the selection and management of network connections.

In the finalised FP6 Gollum project, we implemented a first version of a connection manager able to choose the optimal connection automatically from a range of available networks.

Back to top

1.4Mapping and Location Framework

Mapping and location technologies are being integrated into many different kinds of applications. The trend is to integrate GPS functionality into an ever wider variety of devices. This gives rise to all sorts of possible applications:

• Logistics applications—such as delivering food, gas, etc.—to customers
• Applications targeting tourists—directions to and information about local attractions, museums, resorts, ski areas, theme parks, and cities
• Sales applications, for example, giving appropriate offers depending on the customer's
• Humanitarian services, for example, helping to organize, deploy, and maintain contact with relief workers in cases of natural disaster
• Operating system services like networking systems that can use location information to choose networks

Up to now, however, technologies to aid the development of mapping- and location-aware applications have always been based on proprietary platforms. These proprietary platforms have always come with their own content, and the developer has always been locked into the content of the chosen platform without the ability to create his or her own content or use other third-party maps freely.

EMIC's mapping and location framework provides a programming model that enables application developers to integrate mapping and location information into their applications for mobile devices by providing advanced functionality such as map caching, sign-post for off-screen point-of-interest, and the support for non-linearly scaled maps (e.g. subway maps). We envision mobile devices ranging from portable PCs—like a tablet PC or a normal notebook computer—down to Windows Mobile-powered Personal Digital Assistants (PDAs) and smartphones.

Back to top

2Home

Our researchers focusing on the home environment are currently working on:

• Programming framework
• Recommender systems

Back to top

2.1HomeFx – the programming framework for the connected home

Home technologies span multiple domains, including entertainment, home control, home automation, home security, home networking, and device-to-PC connectivity. In the entertainment and home control domains, the technologies have been very self-contained, are often very proprietary, and can be very expensive. These technologies generally work well for the specific use cases they're designed to solve, but, traditionally, they have been difficult to connect together to solve user scenarios beyond their original scope.

We see the home as continuing to be a heterogeneous system with multiple types of devices that use varied underlying technologies. A connected home is a home in which several devices are connected using a network. The end users in the home need to be able to use and manage those devices in an integrated fashion and much more simply than can be done today. For example, if a user brings home a new Web camera, the home network should be able to recognize the new device, ask and receive authorization from the user to connect the device to the home network, make the connection, and thus allow other devices (PCs, TVs, PDAs) capable of starting to use the services the device brings to the home—in this case, video capabilities—without having to configure each device individually.

The mission of our home team is to develop the concepts and a prototype implementation for a framework that delivers basic functionality for the connected home. This includes an easy and intuitive programming model for UPnP (Universal Plug and Play) devices and Web services in the home, integrated security, automatic device and service discovery, and installation and updates of distributed applications.

Our current cooperation projects include IST FP6 AMIGO (Ambient Intelligence in the Home) and HomePlane (Next Generation Multimedia).

Back to top

2.2Recommender systems

Have you ever sat down to watch TV only to find there isn't anything of interest on? Even if you have hundreds of TV channels from a satellite or cable system, this can be a problem, and when you have hundreds of choices, finding something becomes even more problematic. Combine this problem from traditional broadcast media with the growing amount of multimedia content on the Internet and self-created content, and the information overload problem seems to be getting worse and worse. How do you find content that you're interested in? Isn't there an easier way? We think there is, and we are very interested in discovering new ways to solve this problem using the technology of recommender systems. Recommender systems build an understanding of your preferences and interests and use that to recommend entertainment and information that meets your specific preferences.

The aim of our recommender research is to build a Recommender System that can be used as a research platform for ongoing recommender work. The system will use a hybridized approach, where predictions from different algorithms are merged into unified recommendations to provide the best results. The algorithm engines will be able to plug into both the system, as well as the content databases. This will facilitate integration of developers' algorithms for recommender systems and also allow them to add their own content databases. The personalization will be based on explicit and implicit user behavior—i.e. users will be able to customize preferences directly, but also their actual behavior will be taken into account. We intend to release this system to the community for research use.

A second aim of our recommender research is to develop an easy way to share recommendations and preferences. Allowing users to create genre communities—social networks around media—or simply to share with friends-of-friends' circles is a powerful way to enhance a recommender system. Creating an infrastructure that enables this will foster creation of new applications.

Back to top

3Enterprise

Our enterprise researchers are currently working on:

• Web Services Security and Privacy
• Service Level Agreements
• Workflow
• Qualified Signatures

Back to top

3.1Web Services Security and Privacy

Many security technologies today are reliable in products from Microsoft and others. Though there is still much to improve in these technologies, we see growing problems in managing security systems in an open environment, especially when organizational boundaries are crossed. We therefore concentrate our research efforts in this area:

• Generally, security bootstrapping is still a manual task. It is, for example, hard to set up new "federations" and change security rights. Static and closed environments, therefore, tend to be created.
• Furthermore, security is managed "by hand." The "low-level" security configuration is detached from the "high-level" business and individual security requirements, with the result that the actual implementation often does not match the needed requirements.
• In cases where fine-granular access control is implemented to address rich security requirements, current technologies can only offer giant access rights tables. This typically results in static setups that IT departments resist changing, and actual rights are therefore often old and insecure.

Organizations these days increasingly use Web services for cross-organizational interactions, although typically only for "long-term" relationships and connecting static processes. So-called "federations" (i.e. multiple organizational domains that have established trust with each other) can only be manually set up and require "unbounded" trust relationships, often making them a concern for IT departments.

For dynamic collaborations between individuals of different organizations, technologies other than Web services federations are thus currently preferred. Examples of these include creating vendor accounts, leveraging externally hosted services, using peer-to-peer file sharing, or simply relying on e-mail. Problems associated with these are more complex identity management, separate and stale user accounts, and uncontrolled information exchange.

The mission of our security and advanced privacy team is to enable individuals from different organizations—and more generally, across different trust boundaries—to consume and expose Web service resources in a dynamic and secure way:

• Dynamic: setup sharing and collaborations on the fly, across organizational boundaries, with minimal out-of-band overhead for administrators or users
• Secure: ensure security and privacy requirements of all stakeholders are addressed
• Across trust boundaries, with focus on enterprise, but more generally applicable

Our envisaged strategy for the security side is to provide prototype software components that enable dynamic Web services security on Microsoft platforms to facilitate cross-organization collaboration or other interaction. Particularly, in the context of Web services federation, our work is relevant because it enables more dynamic and scoped federations, where the current products (Microsoft ADFS and others) focus on more static and organization-wide federations. Individuals will be able to set up new federations—for technical support issues, sales presentations, and one-time interactions, for example)—and only expose specific resources and include specific users. These scoped federations will be dynamically created and managed by the collaboration owners themselves, requiring only minimal out-of-band overhead for IT-administrators, yet make it more secure by making it more manageable and supporting central visibility and control.

Of equal concern to our research team is our work on advanced privacy. Privacy is an important personal security requirement of every individual participating in our interconnected information society. Privacy-preserving identity systems are, therefore, a major component of our work. Even more so than for security, privacy is important in scenarios going beyond the enterprise domain.

TrustCoM, MOSQUITO, NextGRID, MYCAREVENT, BREIN, and FIDIS are the current European projects advancing and providing the use cases for our work in this area.

Back to top

3.2Service Level Agreements

Today, more and more applications are delivered through services. Service-Oriented Architecture (SOA) is very valuable for loosely coupling different parts of a system. It adds agility in responding to environmental evolution and helps to provide new functionalities quickly. But to add business viability, it is important to describe the expected business level objectives. A Service Level Agreement (SLA) is a contract that links the service provider and service consumer on the business level objectives of the service. It describes what the service offers and under what conditions the service is offered. It generally includes the description of the two parties, the price, the time validity of the agreement, and the conditions of the agreement. It is important for businesses to monitor the service usage and raise an alert if an SLA violation occurs. The SLA is business critical and has to be enforced automatically. For this reason, the SLA has to be machine readable. This means an SLA has to be transformed into a machine-readable form usable by a program that monitors services and reacts to SLA rules.

The aim of our SLA team is to build prototype software that supports SLA monitoring and auditing. This prototype covers different aspects of SLAs—business objectives and environment, for example. Because SLAs can be used in every scenario where services are involved, it ties into all our other research domains.

SeCSE and Nextgrid are examples of European projects where we work together with partners on SLA technology.

Back to top

3.3Workflow

A workflow describes the operational aspect of a work procedure: how activities are structured, who performs them, what their relative order is, how they are synchronized, how data flows, and how activities are tracked. It is common practice to describe a business as a workflow in a company.

But most workflows are statically defined. When dealing with services composition, workflows define how services are aggregated and integrated. In such a service-rich environment, it is possible multiple copies of a service may exist—for example, multiple semantically equivalent versions of a service on different machines, with different implementations (platform, language, algorithm). Selecting an optimal service from this set of equivalent services is currently a decision undertaken manually by the user. To add flexibility and adapt to SOA requirements of value added services, workflows need to become more dynamic and automatic. This advanced workflow feature is called "Dynamic Selection."

Another aspect of workflows is their complexities and need for complex exception handlings. Workflow engine providers have to look for alternative ways to deal with exceptions, especially in dynamic or highly dynamic scenarios in which a workflow designer is unable to forecast all possible failures. The failing service could help define the strategy needed to recover. This advanced workflow feature is called "dynamic recovery."

The aim of our workflow team is to provide a prototype that enables the execution of dynamic workflows. This prototype should be able to show dynamic service selection from activities in a well-known workflow and dynamic service recovery.

inContext is one European project where we work together with partners on dynamic workflow approaches.

Back to top

3.4Qualified Signatures

In 1999, the European Commission published directive No. 1999/93/EC, which stated the need for EU member countries to implement national standards and regulations for legally binding electronic signatures. Conventional digital signatures appeared to be insufficient to fulfill this need because of a lack of features such as long-term signature validation.

In response to the EC directive, the European Telecommunications Standard Institute (ETSI) developed the XML Advanced Electronic Signatures standard (XAdES). XAdES extends the IETF/W3C XML Digital Signature Standard (XMLDSIG) with additional properties aimed at supporting the EC directive. These extensions (called "Qualified Properties" in XAdES) facilitate the evaluation of a signature by an auditor/arbiter independent of whether the evaluation happens shortly or long after the signing operation. After publication by ETSI, XAdES was submitted to the W3C's XMLDSIG Working Group and was accepted as an official W3C Note commenting on XMLDSIG in February 2003. The XAdES submission to W3C can be retrieved at: http://www.w3.org/TR/XAdES/.

We are continuing to work, together with our partners, on the implementation and verification of the XAdES standard in the European eGovBus project.

Back to top