Problem

1
Department of Computer Science Engineering
M2 Middleware Support for Collaborative
Adaptation
Zhenxiao Yang, Zhinan Zhou, and S. Masoud
Sadjadi Advisors Prof. Betty H.C. Cheng and
Prof. Philip K. McKinley Software Engineering and
Network Systems (SENS) Laboratory

1. The framework

Software systems need to adapt their behavior at
run time in response to a changing environment.
Many techniques have been proposed to address
this issue. The M2 framework provides
communication and collaboration support for
existing adaptation techniques to work together.
M2 and Adaptive Systems
M2 Architecture
M2 Collaboration Protocol

• Introduction
• Problem
• Software needs to change its behavior at run time
  to respond to a changing environment and changing
  requirements
• Many techniques have been proposed to address
  run-time software adaptation
• These techniques need to work together
• Challenges
• There exist techniques with different objectives,
  designed for various platforms, with different
  techniques from various parties.
• Solution
• M2 enables different existing adaptation
  techniques to work together through
• a communication protocol
• an adaptation collaboration protocol
• a set of domain-specific adaptation protocols
• M2 framework
• addresses the collaboration problem among
  multiple applications and multiple system layers
  in adaptive systems

M2 Communication

• The M2 micro-kernel and services
• Applications and platform brokers

• M2 is in the middleware layer
• M2 services elements from all layers

• Star-shaped local message exchange
• Router-router message exchange

1. A Case Study

Result
System Setup
Decision Maker
Collaborative Adaptation Middleware
Heterogeneous Physical Networks

• System adaptation goal
• total system running time ? 250min (given the
  limited battery power resource)
• Application components
• Adaptive video conferencing component
• Adaptive audio conferencing component
• Adaptive textual messaging component
• Platform brokers
• Power-sensor
• detects critical system power levels and alerts
  interested parties
• Services
• Rule-based goal-oriented decision maker
• Administration console

Admin Console
Power Sensor

• By adapting the structure and behavior of the
  system as it executes, we were able to extend the
  lifetime of a Toughbook system by 50 minutes to
  achieve the system running time goal, with an
  acceptable user satisfaction rate.

Acknowledgements This work has been supported
in part by the following grants NSF EIA-0000433,
CDA-9700732, CCR-9901017, EIA-0130724, Department
of the Navy, and Office of Naval Research under
Grant No. N00014-01-1-0744.