Vision Synopsis Projects Testbed Funding Service Guests

Mobile Communications Research

Microsoft Research
One Microsoft Way, Redmond, WA 98052-6399

July 1997 - Present
(last partial update: July 2005)

Our objective is to enhance wireless functionality in the local area and
to push local area wireless system performance and functionality into the wide area.

We are exploring new ways of enabling and exploiting wireless connectivity. In addition, we are working to improve the effectiveness of current and future generation wireless networking and mobile computing technologies.

We are designing software that enables quick-and-easy information exchange between diverse set of wireless devices, and are building a foundation for creating applications and services that are both mobile-aware and location-aware. Our systems are designed to adapt dynamically to the hostile radio environment and to the changing network topology. Our projects emphasizes rapid prototyping and scenario based design so that end-users, developers, and researchers can experiment with actual systems

Projects and Activities

We have several projects underway, in different stages of completion, that are motivated by our overall goal. Here are descriptions of a few:

  1. Self Organizing Neighborhood Wireless Mesh Networks (2002-05)

    Details on

    Several of us in MSR Redmond, Cambridge, and Silicon Valley are working on creating wireless technologies that allows neighbors to connect their home networks. The technology enables inexpensive residential i broadband access in hard to reach areas (rural communities, developing nations etc.)
    Community multi-hop wireless networks are disruptive to the current broadband Internet access paradigm, which relies on cable and DSL being deployed in individual homes. These are important because they allow free flow of information without any moderation or selective rate control. Compared to the large DSL and cable modem systems that are centrally managed, mesh networking is organic, where everyone in the neighborhood contributes network resources and cooperates. We are researching technical problems that continue to remain challenging despite several decades of research in packet radio networks.

    Publications (Click here for complete list)

    Presentations / Video




    • Researchers from MSR Redmond, MSR Cambridge, and MSR Silicon Valley

  2. UCOM: Exploring the role of Multiple Radios in Short-hop Wireless Networks (2001-03)

    Assuming that the cost of RF Transceivers (radios) will continue to plummet, eventually reaching a price point that is negligible, we argue that two or more radios working collaboratively to perform the same task, in a short-hop wireless network improve the performance and functionality of the system dramatically over traditional single-radio wireless systems that are popular today. In this context, we re-visit some classical problems in wireless networking, including energy management, mobility management, channel failure recovery, capacity enhancement, quality of service, bandwidth management, and location determination

    We show that a systems approach, grounded on three design principles: Design for Separation, Design for Choice, and Design for Flexibility, can alleviate many of the problems prevalent in current Wifi based wireless networks. We corroborate our assertion with results obtained from a multi-radio system that we have built. Our experience leads us to take a position that a multi-radio platform is the right platform to base the future of local area and personal area wireless networking. Creating such a platform has serious implications on both hardware and software design as well as on algorithmic and protocol research. We are exploring some of these issues in this research.


    • P. Bahl, SIGCOMM Future Directions in Networking Architecture Workshop, (August, 2003)
    • P. Bahl, NSF-COST(EU) Networking Research
    • Funding Workshop, Invited Talk (June 2003)
    • P. Bahl, University of Maryland, College Park, USA, Center for Satellite and Hybrid Communication Networks. Dept. of Electrical Engineering (September 6, 2002)
    • P. Bahl, University of Massachusetts, Amherst, Massachusetts, USA, Department of Computer Science (September 5, 2002)
    • E. Shih, SIGMOBILE MobiCom Conference, Atlanta, GA, Wake on Wireless: An Event Driven Energy Saving Strategy for Battery Operated Devices, (September 27, 2002) Slide Deck (pdf, 696 KBytes)
    • P. Bahl, Wake on Wireless (April 4, 2002) Slide Deck (ppt, 3924 KBytes)
    • P. Bahl, A. Adya, Lili Qiu, Univeral CoMmunicator (February 25, 2002)
          - Protocol Control Flow (ppt, 2706 KBytes)
          - Software Design (ppt, 67 KBytes)
          - Last Hop Bandwidth (ppt, 373 KBytes)


  3. Wireless Web - Analysis of Browse and Alert Services (2000-01)

    Over the last decade the cellular phone industry and the World Wide Web have experienced a phenomenal growth as people around the world have embraced these technologies at a remarkable rate. Understandably, there is much excitement around the impending deployment of third (3G) and fourth generation (4G) wireless cellular networks at the core of which are cellular phones and Personal Digital Assistants (PDAs) capable of accessing the Web wirelessly from anywhere at anytime.

    To help sustain the growth of the wireless Internet over the next several years, critical issues that needs to be examined are: performance of the "wireless Web" servers and popularity of the content provided by Internet service companies to users accessing the Web wirelessly via small devices.
    We focus on Web servers that deliver notification and browse services over wireless channels. We analyze the notification traces and browser traces to understand the system load, the type of content that is accessed, and user behavior. Some of our key results are: (a) Notification messages are small, and can fit into a single packet in most cases; (b) The popularity distribution of notification messages follows Zipf-like distribution, and a small number of notification messages account for most message transmissions; (c) There is significant geographical locality in notification content; (d) The set of popular queries in browser log remain relatively stable over time; (e) There is significant variation in the load different users put on the Web servers that provide notification and browse services; (f) There is little correlation between how people use notification and browse services, both in terms of the amount of usage and in terms of the popular content categories. We discuss the implications of our findings for techniques.


    • L. Qiu, SIGCOMM Internet Measurement Workshop, (November 2, 2001), Slide Deck (ppt, 2540 Kbytes)

  4. Wireless Internet Access in Public Places (1999-2001)

    To the best of our knowledge, CHOICE was the first free public area wireless hot-spot network in the world. We ran the network in Crossroad Mall, Bellevue (CROWN) for 2 years (1999-2001). This project was then retired. The Crossroad's mall continues to use our equipment and provide free wireless access. Check it out here

    The Choice Network (Nov. 1999 - Aug. 2001)

    The wireless communications revolution is the ability to connect to the Internet from anywhere at anytime. This revolution is typified by the 3rd generation or 3G wide-area cellular systems which are touted as the wave of the future because of their ability to support data networking at speeds of 2 Mb/sec. But is this sufficient?

    Today, local area wireless networks can provide connectivity at up to 11 Mb/sec, and in the near future they will provide access speeds of up to 54 Mb/sec! Consequently, there is and will continue to be, a large difference in performance between 3G systems and local area wireless networks. As we become accustomed to these higher data rates and as the Internet becomes increasingly multimedia-centric, our appetite for faster connectivity will most likely increase.

    The CHOICE Network is about providing a choice to individuals who want to access the Internet from "almost anywhere". Almost anywhere includes places of congregation or public spaces such as airports, restaurants, libraries, train-stations etc. The CHOICE Network is a hardware-agnostic, software-only technology that can propell the deployment of wireless LANs making them ubiquitous. Thus, individuals presenting the proper credentials can access the Internet at speeds that are much greater than 3G speeds from almost anywhere.

    The underlying language of the Choice Network is the Protocol for Authorization, and Negotiation of Services, or PANS. PANS provides authentication, access, charging, security, compression, and last-hop QoS to users of the Choice Network.

    PANS authentication is global and individual-centric i.e. the user can be anywhere in the world and PANS can securely authenticate his or her credential using a globally available server and database. Once this is done, PANS allows access to the Internet in accordance to a pre-configured policy manager. Security is robust with per-user, varying-length keys and varying encryption algorithms valid for a varying amount of time, all of which lead to a system that is considerably more secure than the Wired Equivalent Privacy (WEP) security available in today's popular WLAN products. Last-hop QoS scheduling is based on the Generalized Processor Sharing discipline, combining centralized class-based scheduling with a propriety distributed weighted fair scheduling mechanism within classes.

    CROWN Deployment

    We have deployed the CHOICE network at the Crossroads Shopping Center in Bellevue, Washington. Dubbed CROWN (for CROssroad's Wireless Network), this network enables everyone equipped with an IEEE 802.11b wireless network card to access the Internet after he or she has authenticated themselves to MS Passport. Additional services such as access to local printers, mall portal site, splash screen of services, list of upcoming activities/event, in-store buddy list, and location guidance will also be offered. The pilot is being used to fine tune PANS and serves as a testbed for research on connectivity and computing in public spaces.

    Demos / Coverage
    Related Work
    • Victor Bahl
    • Srinivasan (Cheenu) Venkatachary (no longer with Microsoft)
    • Anand Balachandran (UCSD Intern, Fall 1999)
    • Allen Miu (MIT Intern, Summer 2000)
    • Stephen Dahl, System Support
    • Paul Hoeffer, Web Support
    • Pierre De Vries (No longer with Microsoft)

  5. RADAR: Location Determination and Services (1998-2000)

    An area of study that has fascinated researchers for close to 10 years has been one of building systems and services that are location-aware. Generally speaking, most of the systems developed to-date have required specialized hardware. The cost of deploying this hardware, usually for the sole purpose of location determination, has been prohibitive compared to the perceived benefits of the system. Consequently, wide-scale adoption of location-aware systems has been limited.

    Inspired by the promise of enabling a new class of applications and of adding value to popular wireless LAN technology, we have built an in-building user location and tracking system. Our approach is different from most previous efforts in that we do not require any specialized hardware but instead our system is built upon an already deployed and fully functional wireless data network.

    RADAR is an in-building location-aware system in which radio-frequency (RF) wireless LAN enabled mobile nodes compute their location and use this information to inform other nodes of their location, and to determine which network resources (e.g., printers, scanners, etc.) are within close proximity. The system consists of a signal processing front-end and a directory services, and a service location back-end.

    The front end operates by recording and processing signal strength information from multiple base stations positioned to provide overlapping coverage in the area of interest. It combines empirical measurements and signal propagation modeling with environmental profiling, mobility modeling, and topographical constraints to locate and track mobile nodes. The back-end uses the location information to query a directory server (via LDAP) to locate nearby network resources. Where no directory server is available a service discovery protocol is used.

    Demo / Coverage


  6. Wireless LAN Programming, Performance, and New Applications (1997-99)

  7. WiLIB (Library for Configuring Wireless Hardware) (June 1997 to July 1999; Retired)

    Radical differences in channel characteristics coupled with end-node mobility make wireless networking sufficiently different from wireline networking. Unfortunately, under the current implementation of most operating systems a local area wireless network is treated as just-another-network and is generally exposed to the higher layer networking protocols, operating system, and applications as an Ethernet technology. In our opinion this is a limited programming paradigm since it makes it difficult for the system to adapt to the changing channel conditions and prevents software vendors from developing adaptive and novel applications.

    We have enhanced the programming interface exposed by the Network Device Interface Specification (NDIS) under Windows. With our extensions the software components are assured of compatibility with all WLAN NICs that support this interface. Similarly, hardware that provides the specified interface (via NDIS miniport driver) is assured that software components are able to locate and access its functionality. This, we feel, will lead to smarter software that: (1) improves hardware utilization and system performance, (2) conserves power, (3) improves last-hop quality of service, (4) manages mobility efficiently, and (5) enables new and exciting network services not possible on wired networks.

    The need for this project arose out of the frustration and pain we experienced in porting mobile-aware and wireless-aware applications over hardware from different vendors. Each vendor had his or her own propriety interface. In addition to the usual ramp-up time and effort required to learn the interface, the programming library was generally provided under restrictive terms and agreement which prevented free and timely publication of research results.

    In addition to the NDIS WLAN extensions we are creating WiLIB, a user-level library, that lets programmers configure wireless hardware dynamically (e.g. moving between ad hoc and infra-structure mode etc.). Once completed, we intend to use WiLIB to carry out research on studying the effects of the RF channel on upper-layer protocols and applications, and on providing information to the networking stack that lets these protocols adapt to the changing environmental conditions.


    • Victor Bahl
    • Gavin Holland (Intern, Summer 1999)
    • Josh Broch (Intern, Summer 1998)
    • Tom Fout (Windows Networking, Operating Systems Division)
    • Stephen Hui (Windows Networking, Operating Systems Division)


  8. Fair Scheduling and Quality of Service in wireless LANs

    This project looks at issues that are associated with providing quality-of-service in broadband networks including ad hoc LANs and multi-hop wireless networks. Two important issues need to be addressed in this context: (a) impact of broadcast nature of wireless medium on design of QoS mechanisms, and (b) applicability of traditional "fairness" definitions in the context of ad hoc networks.


  9. Power Aware Ad Hoc Sensor Communications


    We have developed software that allows wireless nodes to create ad hoc local area networks automatically. This software is being experimented with in public places where we are building and deploying services that exploit this functionality. Examples include: proximity networking, buddy-lists etc.

    Global Hailing Channel

    With the proliferation of numerous wireless devices in the market, certain frequency bands are getting congested and noisy. This is specially true in the case of the 2.4 GHz ISM band which is available all over the world and consequently is highly popular among wireless hardware manufacturers. A substantial percentage of the local-area and personal-area wireless data networking standards are being built on the same frequency band, for example: IEEE 802.11, IEEE 802.11b, HomeRF, Bluetooth, IEEE 802.15, and OpenAir are MAC and PHY standards for wireless devices operating in the 2.4 GHz band. For the past several years there has been a steady increase in equipment and users in this band. If the current popularity of wireless networking continues to explode, it is only inevitable that the spectrum efficienty and system performance of these devices will degrade significantly. This in turn, may result in higher user frustration with wireless devices, which potential could slow the rapid growth of the overall wireless LAN-PAN market that we see today.

    During a summer retreat on Invisible Computing, we proposed the idea of creating a Global Hailing Channel. Our proposal is focussed on improving the radio signal congestion problem, while simultaneously providing a universal discovery and negotiation channel and a PHY Protocol, available to devices all over the world. Multi-modal radios use this channel to exchange connectivity and capability information. A white paper on this subject will be available shortly.

    • Victor Bahl, Microsoft Research, Redmond, Washington
    • William Kaiser, University of California, Los Angeles
    • Robert Mayo, Western Research Lab, Compaq Research


  10. RF Standards

    We realize that in order for wireless technologies to succeed it is important for the different hardware and software manufacturers to agree on a common (protocol) language. We are therefore actively involved in standards-setting organizations. Our primary interest is in wireless local area networking which includes enterprise networking, home networking, and personal area networking