Real-Time Computing and Communication
Many real-time systems, such as Voice over IP, and
military/industrial control systems, demand delay-guaranteed services
to meet timing requirements of their applications. Our work has made
contributions in developing delay guarantee techniques on real-time
computing and communication in various systems, including
differentiated services networks, wireless networks, component-based
systems, and therma-aware systems.
Differentiated Services Networks
The differentiated services (DiffServ) model is
well-known for its advantage of being able to support scalable
Quality-of-Service (QoS) for aggregate traffic in computer networks.
However, there is a fundamental dilemma: on one hand, the DiffServ
model proposes to maintain no flow level information at core routers to
help the system scale up; on the other hand, delay-guaranteed services
naturally require fine-granularity treatments by the system. We solved
this problem by proposing a utilization-based delay guarantee technique
in DiffServ networks. With this new method, we were able to
successfully employ a utilization-based admission control approach,
which does not require explicit delay computation at admission time and
hence is scalable to large systems.
Specifically, we developed the utilization-based delay
guarantee technique in DiffServ networks in the following steps: (a) We
first considered deterministic delay-guaranteed services, i.e.,
applications require hard deadlines; (b) We then extended deterministic
delay-guaranteed services to statistical ones, which significantly
improved the effectiveness of admission control by allowing increased
statistical multiplexing of the network resources; (c) Finally, we
applied our theoretical work into reality and implement a
QoS-provisioning system that could be seamlessly integrated to the
existing Voice-over-IP system. Our work made a substantial impact on
the DiffServ community.
- S. Wang,
D. Xuan, R. Bettati, and W. Zhao, "Providing
Absolute Differentiated Services for Real-Time Applications in
Static-Priority Scheduling Networks," in Proc. IEEE
International Conference on Computer Communications (INFOCOM),
April 2001.
- S. Wang,
D. Xuan, R. Bettati, and W. Zhao, "Providing
Absolute Differentiated Services with Statistical Guarantees in
Static-Priority Scheduling Networks," in Proc. IEEE
Real-Time Technology and Applications Symposium (RTAS), May
2001.
- S. Wang,
D. Xuan, R. Bettati, and W. Zhao, "Differentiated
Services with Statistical Real-Time Guarantees in Static-Priority
Scheduling Networks," in Proc. IEEE Real-Time
Systems Symposium (RTSS), December 2001.
- S. Wang,
D. Xuan, R. Bettati, and W. Zhao, "A
Study of Providing Statistical QoS Guarantees in Differentiated
Services Networks," in Proc. IEEE International
Symposium on Network Computing and Applications (NCA), April
2003.
- S. Wang,
D. Xuan, R. Bettati, and W. Zhao, "Providing
Absolute Differentiated Services for Real-Time Applications in
Static-Priority Scheduling Networks," in IEEE/ACM
Transactions on Networking (TON), Vol. 12, No. 2, pp.
326-339, April 2004. (Journal extension of [1])
- S. Wang,
Z. Mai, W. Magnussen, D. Xuan, and W. Zhao, "Implementation
of QoS-Provisioning System for Voice over IP," in Proc.
IEEE Real-Time Technology and Applications Symposium (RTAS),
December 2002.
- S. Wang,
Z. Mai, D. Xuan, and W. Zhao, "Design and
Implementation of QoS-Provisioning System for Voice over IP,"
in IEEE Transactions on Parallel and Distributed Systems
(TPDS), Vol. 17, No. 3, pp. 276-288, March 2006. (Journal
extension of [6])
Wireless Networks
The convenience of wireless communications has led to
a growing use of wireless networks for both civilian and mission
critical applications, many of which require delay-guaranteed
communications. Wireless networks, however, are substantially different
from their wired counterparts, and technologies developed for wired
networks cannot be directly applicable.
The statistical nature of service provided by wireless
links inherently precludes deterministic delay guarantees. Instead, we
provided statistical delay-guaranteed services in wireless networks. We
proposed a virtual traffic model, which assumed that some virtual
traffic flows exist in addition to real ones. These virtual ones would
consume the full unavailable bandwidth of wireless links and be
assigned a strictly highest priority during scheduling. We observed
that packet delays for real traffic in the original system were
identical to delays in this virtual-traffic model. With this virtual
traffic model, we could directly apply the utilization-based delay
guarantee technique we developed in wired networks before to provide
statistical delay-guaranteed services in wireless networks.
- S. Wang,
R. Nathuji, R. Bettati, and W. Zhao, "Providing
Statistical Delay Guarantees in Wireless Networks," in Proc.
IEEE International Conference on Distributed Computing Systems (ICDCS),
March 2004.
- S. Wang,
R. Nathuji, R. Bettati and W. Zhao, "Real-Time
Guarantees in Wireless Networks," in Resource
Management in Wireless Networking, Mihaela Cardei, Ionut
Cardei and Ding-Zhu Du (Eds.), Kluwer Academic Publishers, January 2005.
Component-based Systems
Component-based systems now serve as an important
platform for developing a new generation of computer software.
Reusability is a key factor that contributes to the great success of
component technology. The reuse of components can lead to shortened
software development cycles and reduced software development costs.
However, all component models lack consideration for reusability of
components in extending to provide delay-guaranteed services.
We developed a real-time component-based system that
maintains the reusability of components while providing
delay-guaranteed services. In the component architecture, we augmented
the functional interfaces and context dependencies with contractually
specified temporal interfaces and explicit time-related context
dependencies, and made each component become what we call a
“real-time component”. We built a component-based
resource overlay, which was composed of resource overlay nodes
– real-time components, to isolate the underlying resource
management from applications. From the perspective of an application
designer, the resource unit is a virtual resource (i.e., a real-time
component) instead of the underlying resource. Based on component-based
resource overlay, we could apply the utilization-based delay guarantee
technique to efficiently and effectively provide delay-guaranteed
services while maintaining the reusability feature.
- S. Wang,
S. Rho, R. Bettati, and W. Zhao, "Toward
Real-Time Component-based Systems," in Proc. IEEE
International Real-time Systems Symposium (RTSS) Work-In-Progress
Session, December 2004.
- S. Wang,
S. Rho, Z. Mai, R. Bettati, and W. Zhao, "Real-Time
Component-based Systems," in Proc. IEEE Real-Time
and Embedded Technology and Applications Symposium (RTAS),
March 2005.
Thermal-aware Systems
As power density in processors has increased exponentially in recent
years, processors are prone to overheating caused by the large energy
consumption. Temperature is becoming one of the big concerns in system
design. Naturally, power management plays a key role in maintaining
temperature. However, the majority of the literature focuses on power
management for the purpose of saving energy, not for maintaining safe
temperature levels. While energy and temperature are closely related,
power control mechanisms and temperature control mechanisms are quite
different. This is due to the fact that energy-aware techniques focus
on dealing with the average energy consumption while temperature-aware
ones focus on handling peak energy consumption.
We are interested in thermal-aware real-time systems, which have two
major constraints: the delay constraint for jobs and the temperature
constraint for the processor. Dynamic speed scaling is one of the major
techniques for power management, which can control the power in the
processor by dynamically changing the speed of the processor. Dynamic
speed scaling allows for a trade-off between the performance metrics of
delay and temperature in temperature-constrained real-time systems: to
meet the delay constraint, we run the processor at a higher speed; to
maintain the safe temperature levels, we run the process at a lower
speed. In our research, we aim to investigate this trade-off under
different dynamic speed scaling techniques.
- S. Wang
and R. Bettati, "Reactive
Speed Control in Temperature-Constrained Real-Time Systems," Best Paper Award,
in Proc. Euromicro Conference on Real-Time Systems (ECRTS),
July 2006.
- S. Wang and R. Bettati, "Reactive
Speed Control in Temperature-Constrained Real-Time Systems,"
to appear in a special issue in the Real-Time Systems Journal (RTSJ),
selected by ECRTS 2006 on May 2006. (Journal
extension of [1])
- S.
Wang
and R. Bettati, "Delay
Analysis in Temperature-Constrained Hard
Real-Time Systems with General Task Arrivals," in Proc.
IEEE Real-Time Systems Symposium (RTSS), December 2006.
Technical Report, tamu-cs-tr-2006-5-3.
|