|
|
Networked Real Time and Embedded Systems Laboratory
Department of
Computer Science
The
|
|||||
|
|
||||||
Resource
Management
|
Integration
|
Dependency Algebra
|
Virtualization
|
FPGAs
|
Cyber-Physical
System Stability and Robustness
Cyber Physical Systems represent the convergence of
computing, communication and intelligence sensing and control of our physical
environment. They are the future of real time embedded systems that are at
the center of modern society’s vital physical infrastructures,
including transportation, manufacturing, communication, and defense systems.
From the perspective of system development and
integration using a mixture of new and reused components, the difficult part
is not functionalities that can be encapsulated by components. The challenge
is often the control of interactive complexity. That is, properties of
complex interactions that cannot be encapsulated by components, such as end
to end timing for distributed monitoring and control under dynamic workloads;
security against both external attacks and internal attacks from compromised
components; system stability against not only random hardware failures but
also residual design and implementation bugs; and safe behaviors even when
the system can no longer function properly.
The design and
development of a system of systems has reached a level of complexity that
cannot be handled properly by existing technologies. Severe cost and schedule
overruns are common during development. Incidents during operations have
reached an unacceptable level.
From a system engineering perspective, the key challenges include:
·
Unexpected interactions: resulting from
syntactically compatible but semantically inconsistent abstractions used by
old and new components; implicit and outdated assumptions about the
environment embedded in the legacy codes; and mutual interferences between
independently developed real time, fault tolerance, and security protocols.
·
System instabilities: when faults and failures
in one component cascade along complex and unexpected interdependency
relations, leading to system-wide failures.
·
Inadequate development infrastructure: due to the lack of
integrated and reusable domain specific reference architectures, tools, and
design patterns with known and parameterized real time, stability, and
security properties. This often leads to the use of many new and unproven
designs in product development, creating complex interactions that lead to
many integration and operational problems down the road. In addition, the
lack of reliable wide-area network support for remote sensing and actuations
makes it difficult to develop distributed cyber-physical systems.
Research Areas
· Robust real time system architecture: Advanced real time resource management protocols; technologies and tools for dependency reduction and well-formed criticality ordering; structurally stable design and implementation to safely use complex or COTS components that cannot be fully verified and tested.
· Predictable sensor and actuator network architecture: Network protocols, middleware, and application development tools for predictable, reliable, robust, and QoS-compliant behavior in unpredictable distributed environments.
· Science of design for system composition: Interface and code annotations technologies that make environmental and semantic assumptions explicit; static and dynamic analysis tools to automatically check component semantic compatibility annotations at integration and invocation times; temporal analysis theory and tools for composability of timing behavior.
·
Infrastructure for
interactive complexity reduction: A set of formally specified and validated
coherent real time, robustness, security and networking protocols. An
integrated set of domain models, reference