 |
The goal of the proposed research is to enable rich wireless data
applications at low cost through application and network driven multi-layer
adaptations of communication protocols and dynamic resource management of
handheld platform architecture.
The proposed research will enable the deployment of media-rich wireless data
services at low cost, both in terms of the communication cost (air time) as well
as in terms of handheld cost. By developing multi-layer adaptive protocol
management techniques which are conscious of the requirements of the data
services and the conditions of the wireless channel, we will enable the network
to deliver richer services to more users, thereby reducing the per-user cost. By
developing application-conscious dynamic management techniques for the wireless
handheld platform, we will enable the support of rich wireless data and protocol
processing tasks with low-cost, low-battery, general-purpose, and
application-scalable handsets.
Future wireless communication networks and appliances will have to support
richer, and more diverse, applications, services, and protocols. Though new
generations of wireless technologies promise to significantly increase the
available bandwidth and data rate, more wireless data subscribers, richer
services, and inherently dynamic wireless network conditions, will mean the need
to manage and better utilize the communication resources effectively. Similarly,
while advances in semiconductor technologies may enable wireless handheld
appliances to deploy more powerful CPU, memory, and battery resources, practical
constraints like cost and heat dissipation, together with the need to support
richer and more demanding protocol and signal processing, mean the appliance
platforms will always be challenged to handle more with less, by managing the
use of its limited resources effectively.
The goal of the proposed research is to enable wireless applications at a low
cost (i.e. energy, service cost etc.) by exploiting configurability in
communication protocols as well as appliance platform architectures. To this
end, we propose: (i) a multi-layer protocol adaptation strategy that exploits
configurability across protocol tasks to satisfy desired application performance
and/or to increase overall network performance; (ii) dynamic platform management
techniques, which allocate and configure the platform resources, at run-time, to
enable their optimal utilization; and by relaying the computational constraints
to the application layer, leading to more effective application level
adaptation.
Our proposed multi-layer protocol adaptation would consist of different adaptive
tasks, spanning across application, link, and physical layer, which will
actively use information from other layers, and a thin vertical layer to
exchange information between layers effectively. In network-wide
application-level services, we plan to investigate how knowledge of the network
links as well as appliance level information can be better utilized to enable
efficient data delivery schemes. In the MAC layer, we plan to develop adaptive
carrier sense multiple access/collision avoidance (CSMA/CA)-based channel access
mechanisms to enable service differentiation to meet individual application
requirements without sacrificing fairness between traffic flows. Similarly, in
the physical layer, we enable adaptivity through the use of multi-direction
antennas and adaptive coding techniques to increase network level throughput. We
also investigate how the proposed physical layer techniques can be effectively
utilized by medium access mechanisms.
To enable the use of low-cost wireless handhelds that are capable of supporting
a rich variety of applications and protocols, we propose the use of
general-purpose configurable platform architectures as opposed to more
customized hardware implementations. We propose to develop dynamic platform
management techniques that detect time-varying application characteristics, and
customize the configurable platform at run-time, to provide performance and
energy consumption characteristics that are more commonly associated with custom
implementations. Dynamic platform management will be implemented in software,
making the addition of new applications to a handset relatively easy and
economical - adding new functions to a deployed platform involves simply
upgrading the platform management layer, along with application software.
Dynamic platform management will also provide applications with feedback about
current platform resource availability. This information can be used by the
applications to regulate the quality of service/content, and thereby enable more
efficient use of platform resources, and hence, a richer suite of concurrent
applications.
The following CWC faculty are participating in this research project: Sujit Dey(lead PI),
Rene Cruz, and Ramesh Rao.
|
|
|
 |
|
Affiliated Faculty
CoRe Research Projects