1. TRUST:Team for Research in Ubiquitous Secure Technologies
Systems Science Challenge Area
Douglas C. Schmidt
Area Coordinator
NSF STC Review
September 13th 2004

2. Systems R&D Accomplishments
TRUST
Systems R&D Accomplishments
Productivity & quality gains from higher-level abstraction mechanisms & tools
Partially automated solutions for limited domains
Analysis
Simulation
Generation 2
NSF STC Review
September 13th 2004

3. New Systems Research Challenges
TRUST
New Systems Research Challenges
Complex Interdependency Modeling & Analysis
Model- based Integration of Secure Systems
Secure Networked Embedded Systems 3
Software Tools for Design & Information Management
NSF STC Review
September 13th 2004

4. Complex Interdependency Modeling & Analysis Challenges & Goals
TRUST
Complex Interdependency Modeling & Analysis Challenges & Goals
Characterize critical dependencies between interconnected networks
e.g., what information should be transferred between the networks?
Devising control architecture of networks to ease the computational burden in each network
e.g., time scale of convergence of state estimation algorithms critically influences the severity of impact of failures
Interdependent
Critical Infrastructure 4
Finding the right trade offs
e.g., privacy, dependability, security, performance, predictability
NSF STC Review
September 13th 2004

5. TRUST
Complex Interdependency Modeling & Analysis Solution Approaches: Robustness from Scale
Although the emerging generation of networks is locally fragile (node failures, channel impairments), the large-scale allows for global robustness 5
Sensor Networks
Ad Hoc Wireless Networks
NSF STC Review
September 13th 2004

6. Model-based Integration of Secure Systems Challenges & Goals
TRUST
Model-based Integration of Secure Systems Challenges & Goals
Domain-Specific Modeling Languages
Multiple-aspect modeling languages for systems/security co- design
Understanding and modeling inter- dependence between security aspects and core systems aspects
Analysis tools for co- verifying security, performance and safety properties
Matlab
Code-Gen.
Matlab
Code-Gen.
Config.
Generator
Model-Driven Generator
Technology
Modeling of generators
Generating generators
Provably correct generators
Embeddable generators
if (inactiveInterval != -1) {
if (thisInterval > inactiveInterval) {
(int)(System.currentTimeMillis() - lastAccessed) / 1000;
int thisInterval =
invalidate(); }
ssm.removeSession(this);
ServerSessionManager ssm =
ServerSessionManager.getManager();
private long lastAccessedTime = creationTime;
* session, as the number of milliseconds since midnight, January 1, 1970
/**
* Return the last time the client sent a request associated with this */
public long getLastAccessedTime() {
* a value associated with the session, do not affect the access time.
* GMT. Actions that your application takes, such as getting or setting
return (this.lastAccessedTime);
this.lastAccessedTime = time;
* should be called by the context when a request comes in for a particular
* Update the accessed time information for this session. This method
this.lastAccessedTime = this.thisAccessedTime;
this.thisAccessedTime = System.currentTimeMillis();
public void access() {
* session, even if the application does not reference it.
lastAccessedTime = ((Long) stream.readObject()).longValue();
lastAccessedTime = 0L;
this.isNew=false;
isNew = ((Boolean) stream.readObject()).booleanValue();
maxInactiveInterval = ((Integer) stream.readObject()).intValue(); 6
Configuration Specification
Code
Analysis Tool
NSF STC Review
September 13th 2004

7. Model-based Integration of Secure Systems Solution Approaches
TRUST
Model-based Integration of Secure Systems Solution Approaches
Access Control
Meta-Model
Composition
Meta-Models
GME
Meta-Modeling
Multiple-aspect modeling languages are defined by formal meta-models
Security models are built independently from platforms and expressed as design patterns
Model Weaving technology is used to generate integrated security/systems models
Model-based generators are used to generate systems on Secure Platforms.
built by
generated
from
Security Models
Composition
Models
GME S-ESML
Modeling
built by
generated from
Integrated
Model
Model Weaver
weaved by
generated
from 7
Secure Platform
NSF STC Review
September 13th 2004

8. Secure Networked Embedded Systems Challenges & Goals
TRUST
Secure Networked Embedded Systems Challenges & Goals
Automated design, verification, & validation: Support simultaneous design & propagation of constraints among different domain-specific design teams
Verified design, in a mathematical or formal sense
Validated design, in an engineering sense
Certifiable design, to allow regulatory agencies to certify the production software
Applications
Applications
Applications
Sensors
Controllers
Actuators 8
Operating
System
Operating
System
Operating
System
Endsystem
Networks
Endsystem
Networks
Endsystem
NSF STC Review
September 13th 2004

9. Secure Networked Embedded Systems Challenges & Goals
TRUST
Secure Networked Embedded Systems Challenges & Goals
Secure, composable, & adaptive software: Build modular middleware services that support secure embedded systems
Support for a variety of performance tradeoffs
In-network processing
Autonomous adaptation to system conditions
Assurance of a high level of security
Differentiate between malicious intrusion & system failures
Applications
Applications
Applications
Sensors
Controllers
Actuators 9
Operating
System
Operating
System
Operating
System
Endsystem
Networks
Endsystem
Networks
Endsystem
NSF STC Review
September 13th 2004

10. TRUST
Secure Networked Embedded Systems Solution Approaches: Integrated Research Tools
Software technology
Middleware & generative techniques to automatically manufacture highly optimized software using high-level design models & domain-specific configuration knowledge
Hardware architectures for embedded sensor networks
Berkeley motes as sensor network devices & asynchronous hardware architectures: Cornell Sensor Network Asynchronous Processor (SNAP)
System support
Integrated system architecture for sensor networks building from extensive work on TinyOS & MagnetOS to support secure, reliable, self-configuring sensor networks
Applications
Applications
Applications
Sensors
Controllers
Actuators
Integrated
Middleware,
Operating
Systems,
Protocols, &
Hardware
Integrated
Middleware,
Operating
Systems,
Protocols, &
Hardware
Integrated
Middleware,
Operating
Systems,
Protocols, &
Hardware 10
Endsystem
Networks
Endsystem
Networks
Endsystem
NSF STC Review
September 13th 2004

11. Software Tools for Design & Information Management Challenges & Goals
TRUST
Software Tools for Design & Information Management Challenges & Goals
Software developers depend on complex platforms, & increasingly work by extending or customizing with extra code
Web Services, J2EE, CORBA, .NET
The quality of these platforms and “tools” is a direct determinant of the quality of their applications and solutions
We need technologies to overcome limitations with existing platforms:
Scale poorly & can “melt down” under stress
Are insecure - easy to disrupt or intrude
Are human intensive to deploy, configure
Are hard to repair when disruption occurs
Are costly to own & operate 11
NSF STC Review
September 13th 2004

12. Software Tools for Design & Information Management Solution Approaches
TRUST
Software Tools for Design & Information Management Solution Approaches
Astrolabe captures system state hierarchically, using P2P protocol that “assembles a puzzle” without any servers
Develop new technologies based on peer-to-peer interaction styles that substitute probabilistic objectives for classic deterministic ones
Apply these technologies to overcome limitations with prevailing reliable client/server model, which imposes O(N) delays & O(N2) performance degradation
Name
Avg Load
WL contact
SMTP contact SF
2.6
ITH
1.8
Paris
3.1
SQL query
“summarizes” data
Name
Load
Weblogic?
SMTP?
Word Version
gazelle
1.7
4.5
zebra
3.2 1
6.2
gnu .5
Ithaca 12
Name
Load
Weblogic?
SMTP?
Word Version
swift
2.0 1
6.2
falcon
1.5
4.1
cardinal
4.5
6.0
San
Francisco
NSF STC Review
September 13th 2004