Adaptive Middleware for Heterogeneous Dynamic Environments

1
Adaptive Middleware for Heterogeneous Dynamic
Environments

• Nalini Venkatasubramanian
• Distributed Systems Middleware Group
• School of Information and Computer Science
• University of California, Irvine

2
QoS Aware Information Infrastructure
Electronic Commerce
Adaptive Middleware To Support ilities
Requirements - Availability, Reliability,
Quality-of-Service, Cost-effectiveness, Security
3
The UCI Contessa Team

• Faculty
• Nalini Venkatasubramanian
• Doctoral Students
• Sebastian Gutierrez-Nolasco
• Qi Han
• Yun Huang
• Shivajit Mohapatra
• Jehan Wickramasuriya
• Mayur Deshpande (RESCUE)
• Suraj Nagasrinivasa (RESCUE)
• Bo Xing
• Xing Zheng

• Masters Students
• Bhaskar Chatterjee
• Mark Panahi
• Henry Yan
• Xiaoping Wei
• Humeyra Altintas
• Daniel Massaguer
• Undergraduate Students
• Christopher Bell
• Jason Collins
• Sadullah Nader
• Sarvesh Regmi
• Preethi Srinagesh
• Michael Philpott
• Visitors
• Marco Domenichini (Univ. of Bologna, Italy)
• Luca Clementi (Univ. of Bologna, Italy)
• Alfredo Weitzenfeld (ITAM, Mexico)

4
UCI Artifacts and Achievements

• Research results dissemination
• 30 published/accepted papers, 7 in
  submission/revision
• System Artifacts
• 5 prototype systems developed
• Visibility
• Invited talks and tutorials
• UCI PI on 30 international PCs
• UCI Contessa Students on PC and organizing
  committees

5
UCI Research Thrust Adaptive and Composable
Middleware
Design and develop safe adaptive middleware for
distributed applications
Genetic Algorithms
Game Theory
When, where, how to adapt
Formal Methods Foundation
Machine Learning
Algorithms
Statistical Methods
Systems
Arsenal
Design, implementation, evaluation
Graph Algorithms
MAUDE
6
Research Collaboration

• Contessa Collaborations (ongoing and upcoming)
• Adaptive Secure Group Communication (UCI/UIUC)
• Support for DOS resilience (UCI/UIUC/UPenn)
• Heterogeneous Interoperability (UCI/WUSTL)
• Modeling and managing Sensor networks
  (UCI/WUSTL/UIUC)
• Power-aware mobility (UCI/UPenn)
• Collaborations with other institutions
• UCSD, SRI, IBM
• Interesting External Collaborations (Upcoming)
• NASA
• SPAWAR

7
CONTESSA Technologies
Probabilistic Formal Models
Formal Prototyping Analysis
Mobility Context-Awareness
Formal Prototyping Analysis
DoS Resistant Secure Communication in
Heterogeneous Networks
Adaptive and Context-Aware Secure Communication
in Mobile Environments
DoS Resistant Protocols
DoS Resistant Secure Tunnels
Power-Awareness
Adaptive Reflective Middleware
8
Highlighted Research Context Sensitive
Adaptation

• Context Collection Adaptation in Mobile
  Environments
• Adaptive Data Collection in Heterogeneous
  Environments
• Adaptive Secure Group Communication
• Power-Aware Reconfigurable Middleware
• Systems Support for Heterogeneous Interoperability

Focus topic
9
Running Theme C2 (Composability
Customizability)

• Safe Composability
• Protocol Interference
• E.g. reliable, secure communication
• Protocol Service Interactions
• E.g. secure mobility
• Safe Customizability
• Relaxed notions of safety to accommodate
  uncertainty and ensure performance

10
Research Approach

• Formal Methods to ensure correctness
• TLAM ( A Two Level Meta-architectural model)
• A Structured approach to reflective middleware
• Russian Dolls (UIUC/SRI collaboration)
• Support for interoperable reflection
• Prototype Systems for validation
• CompOSEQ - Safe Reflective Middleware

11
Adaptive Context Information Collection

• Challenges
• Competing goals of Quality of Service
  (timeliness, reliability), Quality of Data (data
  accuracy) and collection overhead
• Our Approaches
• Steer Data Representations
• Design algorithms to support diverse application
  requirements
• Mobile, real-time, sensor
• Prototype implementation
• AutoSeC (Automatic Service Composition)

DATA FLOW
QUERY FLOW
client cache
server
server cache and archive
producer its cache
12
Adaptive Secure Group Communication ( A
UCI/UIUC/SRI Collaboration)

• Mobility brings new needs
• Adaptivity
• Highly dynamic membership changes
• Changing resources (device, network)
• Location awareness
• Our Approach
• Built on relaxed synchrony semantics
• Design adaptation techniques
• Formal prototyping using Maude
• Allows us to experiment with an abstract
  mathematical but executable specification of the
  system

13
Adaptive Secure Group CommunicationProviding
Security with Relaxed Synchrony

• Preserve properties
• backward secrecy and EVS semantics
• Adaptations for enhanced performance
• Key caching
• Lazy key establishment
• Key caching and lazy key can be safely composed

14
Adaptive Secure Group CommunicationNew Degrees
of Flexibility

• Secure delivery modes
• Non-secure
• Clear text
• Secure
• Any (possibly old) known key else nondecryptable
• Ultra-secure
• Most recent known key else nondecryptable
• Secure safe
• Any (old) key, deliver only if all members
  receive and decrypt with same key
• Ultra-secure safe
• most recent key, deliver only if all members
  receive/decrypt with most recent key

Relaxing both synchrony and security have opened
new possibilities !
15
Power- Aware Reconfigurable Middleware (DYNAMO)

• To build a power-cognizant distributed middleware
  framework that can
• exploit global changes (network congestion,
  system loads, mobility patterns)
• co-ordinate power management strategies at
  different levels
• (application, middleware, network, OS,
  architecture)
• maximize the utility (application QoS, power
  savings) of a low-power device.

Directory Service
network
Proxy
GLOBAL PROXY BASED ADAPTATION
16
Focus Topic Multi-dimensional Interoperability

• Future Mobile applications
• Higher resource needs (bandwidth, battery)
• Need ubiquitous service availability
• Approach
• Exploring various forms of heterogeneity and
  develop systems that interoperate under varying
  conditions
• Network heterogeneity
• Wireless modes (ad-hoc and nomadic), Wired LAN,
  Distributed WAN
• Computing Platform Heterogeneity
• Client-Server, Mobile Agents, Grid Computing and
  P2P Services
• Data Heterogeneity
• Real-time video, audio, images, sensor data,
  compute intensive tasks

17
Application Scenario

• Mobile hosts (in ad hoc groups)
• need computational or data intensive services
• Objective Seamlessly bridge the connection
  between mobile hosts and resources so that
• high quality data when connected to access point
• Lower quality information from neighbors o/w

Bring the power of distributed idle resources to
a constrained mobile device
18
Multidimensional Interoperability Demo(A
UCI/WUSTL collaboration)
19
MAPGrid A New ParadigmGrid Enabled Mobile
Applications
Broker
20
MAPGrid Research Challenges

• Resource Discovery
• Define an optimal grid resource allocation
  given
• Changing location of mobile host
• resource heterogeneity of grid resources
  (volunteer servers)
• Varying availability of grid resources (volunteer
  servers)
• Adapt to changing context (mobility, energy)
• Ensure user QoS
• Data Management
• Guarantee data availability
• Resource utilization considering dynamic user
  connectivity

21
Grid Resource Discovery for Mobile Services
Partition Service Period
Volunteer Server Allocation

• Problem
• Determine number of partitions and size of each
  partition
• Machine learning approach using mobility info

All chunks have VS assignments?
N
Y
Rescheduling if dynamic changes happen
22
Grid Resource Discovery for Mobile Services
Partition Service Period
Volunteer Server Allocation
All chunks have VS assignments?
N

• Optimizing grid resource selection
• Factors Load, location, availability, capacity
• Achieve system-wide load balancing
• Maximize the number of accepted services
• Graph theoretic approach

Y
Rescheduling if dynamic changes happen
23
Grid Service Discovery (Multimedia Application)
Collections of services support by grid
significantly increases request acceptance and
Completion ratios
P2 Deterministic data placement T2 Random VS
availability
24
MAPGrid Prototype
Failure detection for both client and VSs using
Globus life time management control.
Volunteer Servers
GridFTP
Globus Toolkit
Information Services
Broker
Life time control
JADE
Http
Data
The Volunteer Server can join and disjoin the
grid dynamically .
Client can specify QoS requirements via friendly
GUI
Client
JMF (Java Media Framework)
25
Systems Support for Heterogeneous
Interoperability
Future
Ongoing Contessa Research
Basic Support for Composable Services
26
Future Cross Layer Interoperable Adaptation

• Challenge
• Support reliable secure real-time QoS
  across multiple communication technologies
• Proposed Research
• Always Best Connected
• Cellular, WiFi, Bluetooth
• Application aware bandwidth sharing and access
  selection
• Cross layer adaptation
• Quality/power tradeoffs
• Admission Control
• Content Adaptation
• Mobile Groups in an ABC world

WLAN, LAN
GPRS/UMTS cdma2000
PAN Bluetooth
DSL, Cable TV
27
Future Adaptation in Pervasive Spaces

• Challenge
• Capturing, processing, delivery and management of
  information in pervasive spaces
• Proposed Research
• Multisensor Location Awareness
• Spatiotemporal Service Composition
• Network Management in Large Multisensor
  environments

28
Summary

• Adaptive Middleware Support for
• Adaptive Context Collection
• Secure, Reliable, Mobile Groups
• Power-Aware Reconfigurable Middleware
• Adaptive Middleware for Multidimensional
  Interoperability
• Future
• Adaptive Interoperability in Pervasive
  Environments
• Formal methods for cross layer adaptations
• Tradeoffs Security, Reliability, QoS