MIT Department of Electrical Engineering & Computer Science

E E C S

Architecture and Design of Mobile Multimedia Systems

Dr. Mani B. Srivastava
Bell Laboratories

Thursday, April 18, 1996
9:00 AM (8:45 refreshments)
Grier Room, 34-401A
EECS Special Seminar

Abstract

The vision of "anytime anywhere anyform" information access is driving conventional systems towards networked computing and communication systems that will provide mobile users with multimedia services such as audio/video conferencing and messaging, video database access, WWW browsing, and live sensor feeds in hazardous workplaces. System architecture is therefore taking a new meaning, beyond solving conventional problems of CPU design, memory hierarchy etc. Instead, a system architect must focus on non-traditional problems such as wireless access, user and terminal mobility, digital signal processing of continuous media, context adaptive applications, and energy efficiency.

To bridge the gap created by these new problems, my research has addressed them from three perspectives: (i) multimedia-oriented mobile network infrastructure, (ii) mobile multimedia terminals and multimedia servers, and (iii) design techniques for low-power and digital signal processing computation After briefly describing my research in the latter two, my talk shall focus on mobile and multimedia network architecture.

I shall describe my work on the SWAN mobile computing system, developed at Bell Labs with the goal of driving wireless networks towards a multimedia-oriented integrated service model. The dream is to enable multimedia services to mobile users carrying wireless terminals of varying degrees of smartness. SWAN (Seamless Wireless ATM Network) consists of basestations connected by a wired ATM backbone network, and wireless ATM last hops to the user terminals.

The key feature of SWAN is the use of end-to-end mobile and wireless ATM. The virtual circuits in ATM allow service quality guarantees to be given to multimedia connections over the wireless link as well. However, the question of how best to make ATM mobile and wireless needs proper addressing. Making ATM mobile requires one to continually reroute virtual circuits, with minimal disruption, as a mobile terminal moves. I shall describe a novel rerouting approach based on virtual circuit extension coupled with loop removal and mobile initiated performance triggered partial virtual circuit rebuild. Making ATM wireless requires the lower layers to be cognizant of ATM. I shall describe how the medium access control and air-interface subsystems in SWAN, besides their traditional functionality, also cooperate with the higher ATM layers to allocate wireless resources to individual virtual circuits.

A prototype implementation of SWAN has been done using mobile user terminals and basestations based on custom designed wireless ATM adapter cards and off-the-shelf 2.4GHz ISM band radios. The system implementation involved a three-way hardware-firmware-software partitioning trade-off. Using the prototype system, we have investigated both native mode ATM applications (using end-to-end mobile and wireless ATM) and conventional TCP and UDP multimedia applications (using IP-over-wireless-ATM). I shall present performance data from initial experimentation, and explore the relationship between performance and wireless link attributes.

URL of this page: http://www-eecs.mit.edu/AY95-96/events/47.html
Created: Apr 8, 1996 | Modified: Jun 25, 1997
This announcement is from the MIT EECS 1995-96 archive. | Current events
To MIT EECS home page | Your comments and inquiries are welcome.