cyber physical systemsjohanl/educ/ocps/ocps 01 intro.pdf · communication core. ... v2v v2v...
TRANSCRIPT
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 1
an introduction and a perspective
Johan Lukkien
Cyber Physical Systems
What is CPS about?
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 2
CPS: Definitions
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 3
• UC Berkeley (May 2015). “Cyber-
Physical Systems (CPS) are integrations
of computation, networking, and physical
processes.” (looks a bit outdated)
• “Cyber-physical systems (CPS) are
next-generation embedded systems
featuring a tight integration of
computational and physical elements.” Annotated concept map from
http://cyberphysicalsystems.org/
ICCPS @ April 2013
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 4
• “Cyber-physical systems
(CPS) are physical and
engineered systems whose
operations are monitored,
coordinated, controlled and
integrated by a computing and
communication core.
• This intimate coupling between the
cyber and physical will be manifested
from the nano-world to large-scale
wide-area systems of systems.
• And at multiple time-scales.”
ICCPS @ May2016
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 5
• “Cyber-physical systems
(CPS) are physical and
engineered systems whose
operations are monitored,
coordinated, controlled and
integrated by (such) computing and communication.
• Broad deployment of cyber-physical systems is
transforming how we interact with the physical world as
profoundly as the world wide web transformed how we
interact with one another, and harnessing their
capabilities holds the possibility of enormous societal and
economic impact.
Some more…
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 6
• “A cyber-physical application is a computer system that processes and
reacts to data from external stimuli from the physical world and make
decisions that also impact the physical world” – J. Sztipanovits. Composition of Cyber-Physical Systems. In: Engineering of Computer-
Based Systems, 2007. ECBS’07
(cited from White et.al;, R&D challenges and solutions for mobile cyber-physical
applications and supporting Internet services, Journal of Internet Services and
Applications, May 2010, Volume 1, Issue 1, pp 45-56)
• “A cyber-physical system (CPS) is a system featuring a tight
combination of, and coordination between, the system’s computational
and physical elements.” – Wikipedia, May 2013
• A CPS is therefore a system composed of physical entities such as
mechanisms controlled or monitored by computer-based algorithms. – Wikipedia, May 2016
So, what is CPS about?
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 7
• Eq. 1: CPS = Computing + Control
– (“Embedded 2.0”)
• Eq. 2: CPS = Embedded Systems + Internet of Things
– (“Embedded meets Ubiquitous Sensing and Communication”)
• Eq. 3: CPS = Embedded Systems + Internet Services
– (“Embedded meets Big Data”, “Virtual embedding”)
• Eq. 4: CPS =
Physical world + Sense/Act. + Control + Computation + Communication + Data
• The novelty is to consider them all together
– a theory of CPS, in which 6 aspects influence each other, are collectively
optimized, enable new applications
– (some) interactions go through the physical world
embedded systems hybrid systems control theory
Using Data 1. Improve subsystem operation
– increased knowledge of the system context from external sources
• weather, environment, …
– evaluate effects of actuation and learn to improve
2. Supply detailed, local data to enable new applications (or improve
existing ones), e.g.,
– internal details of production processes
– wear and tear of internal components
3. Take (increasingly larger) data processing in control loops
– increases in sensor numbers, sampling frequency and sample size
4. (Some) control in the cloud, e.g. cloud based robotics
5. In the end, a CPS ‘knows’ the world beyond sensing and actuating
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 8
Can you give an example?
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 9
Intelligent Transportation Systems
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 10
Vehicles operate using networked ICT
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 11
Local
Control
Local
Control
Local
Control
In-car network
Driver
Control
In-vehicle networks
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 12
• Networks of ECUs
– 40-80 in a modern car
• Designed for
– cooperative behavior
– specialist (remote) management /
diagnostics
• Gateway support for isolation
K-CAN
System
MOST K-CAN
Periphery
SI-BUS
(Byteflight)
PT-CAN
Diagnose
Gateway
BMW 7 series infrastructure
Vehicles become parts of a larger whole
• Vehicle to Vehicle
(V2V) communication
– ad-hoc
– collaborative
applications
• Applications are
distributed
• Actuation may pass
by driver control
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 13
Driver
Control
Driver
Control
Driver
Control
V2V network
Accident prevention
Multiple technologies
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 14
Access Network
GW
RSU RSU
Node Server
OBU
AU
OBU
AU
OBU
AU
HS
Internet
AU Application Unit
OBU On Board Unit
GW Gateway
RSU Road Side Unit
HS Hot spot
In-Vehicle
Domain
Infrastructure
Domain
Ad-Hoc Domain
C2C-CC (Car2Car Communication Consortium initiated by six European car manufacturers) architecture (~2010)
IEEE 802.11p IEEE 802.11a/b/g Other wireless technology (full coverage)
A conceptual view
15-Sep-16
15
V2V V2V
Internet, V2I
Congestion control Road maintenance Environment control End-user applications Car sharing Event management
Driver Control
Driver Control
V2V network
Accident prevention
Local Control
Local Control
Local Control
In-car network
Driver Control
(2)
(1)
(4)
(3) • Example data flows:
– (1) gather detailed driving data to
determine
• local weather
• road condition
– (2) accident prevention by direct
intervention
– (3),(4) informing driver
about upcoming road conditions
Other examples (from Dutch rCPS project)
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 16
medical
imaging
chip
fabrication
professional printing
platoonin
g
server farms
subcutaneous sensing
electron
microsco
py
How did CPS evolve?
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 17
CPS is a convergence
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 18
1. High-performance (embedded) real-time control systems
• Integration of control, computation, communication
• Focus on dependability, robustness,
strict (timing) guarantees
2. Deep penetration of sensing and actuation into
the physical world
• Internet of Things – unified protocol and addressing scheme between
any pair of (electronically enriched) objects
– large-scale heterogeneous data processing
• Focus on scalability, connectivity
• Uncertainty as essential ingredient (wireless, unreliable components)
From: Network QoS Management in
Cyber-Physical Systems, Feng Xia et. al
picture by Maurice Heemels
High Performance (embedded) real-time control
15-Sep-16
19
applications
platform
network
Reservations, Budgets
Admission test
(temporal) Isolation
Real-time interfaces
Local
Control
Local
Control
Local
Control
In-vehicle network
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking
Internet of things
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 20
• A unified protocol and naming
scheme between every pair of
devices
• Pervasive, extending network
communication to billions of
endpoints
• Reaching into the physical
world, reaching deep into
(production) systems, gathering
large amounts of detailed
information about states and
events
Single function devices,
sensors, actuators, M2M, 1-many
electronics, ~100B
Laptops, desktops,
phones, 1-1 electronics
~B
Internet ‘Core’,
Web, Data,
Compute Servers
~M
Two worlds
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 21
• Pervasive connectivity moves into the
safety critical domain
– by including actuation
– by penetrating safety critical systems
– uncertainty and concerns of connectivity
and scalability are complemented with
timeliness and dependability
• Safety critical systems (CPS) are
becoming connected
– by including open (and wireless) networks
– robustness and (timing) guarantees are
complemented with scalability and
uncertainty
How are these systems integrated?
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 22
System of systems
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 23
applications
platform
network
Connected domains
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 24
• Applications run on top of
connected (managerial)
domains
– an application takes
resources from that domain
and possibly runs code
– within a domain a single
behavioral policy may be
assumed
– the overall application is
emergent
• The perimeter is not that clear
– the domain can be more
logical than physical
• IoT example: smartphone apps
– take resources from phone,
network, back office cloud
– crossing of managerial domains
by user consent
• sometimes policy, sometimes
just yes/no
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 25
Intelligent Transportation
Systems
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 26
• Vehicle: integrated system components
– systems by themselves, but not
independent
• V2V: system of systems (vehicles) with
V2V applications
– e.g. accident prevention, parking spot
finding, collaborative driving
• V2I: ‘slower’ (higher latency)
applications, global applications
• ITS is a SoS and a typical Cyber
Physical system
Smart grid
Water supply
Industry
Mobility management
Public services
Smart City
Emergency
Well-being
Navigation
Safety
Applic
ations Management of
carbon release
Energy load balancing
Street Lighting
Connected light poles
Outdoor sensing
Surveillance infra
structure
Smart Mobility
Traffic lights
Road side units
Connected vehicles
Telecom
Cellular network
Data services
Smart Building
IT network
Building management infrastructure
Connected luminaires
Wireless sensors
Smart Energy
Smart meter
Renewable energy sources
Another
typical SoS
SoS characteristics: literature
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 29
• Operational Independence
• autonomous behavior and goal
of subsystems
• Managerial Independence
• subsystems managed by
different authorities
• Evolutionary Independence
• Geographic distribution
• Emergent behavior
• properties deriving from the
combination of subsystems
• properties difficult or impossible
to deduce from subsystems
• Heterogeneity
• Networks as integration point
Consequences of composition
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 30
competition of
control - local versus global
architectural
diversity uncorrelated
requirements
Sharing leads to mixed criticality
• Platforms run applications (components) with different criticality
• Components may have an essential and an optional part
• High demands –worst case – might be rare
• Mixed Criticality
– a mixed-critical system is an integrated suite of hardware, operating
system and middleware services and application software that supports
the execution of safety-critical, mission-critical, and non-critical software
within a single, secure compute platform (from http://www.cse.wustl.edu, research agenda for Mixed-Criticality Systems)
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 31
How to engineer (compose) SoS applications?
• E.g. a Smart City application
• Lewis et al.: three step life cycle
1. list all software services in the
concerned subsystems
2. build the integrated SoS
application from these
3. examine and realize the
consequences for the original
subsystems
• e.g. access to data, using
computation and data resources
• Some thing like this has to be
done but
– this assumes a stable
software ‘base’ (independent
evolution!)
– it requires subsystem
managers to be involved
– it invites adjustments of
subsystems to the applications
at hand leading to
maintenance problems • …. reducing dependencies is key
– it is difficult to involve third
party application developers
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 32
Some requirement for a composition
architecture • Separate the following
– development API for application developers (“North side”)
– integration API for subsystems (“South side”)
– adaptation layer for subsystems
• Interfaces include
– reservation of resources (network and others)
– policies, and negotiation thereof
– coordination of SoS applications
• Models are required for
– reasoning about compositions, optimization
– simulations, at varying levels of detail
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 33
INTEGRATION AND COORDINATION
subsystem 2
Adaptation
Policies / Reservation
Control / Data / Code
Application 1 Application 2
subsystem 1
Local App
Service Discovery
Adaptation
Local App
legacy collaboration
Discovery (subsystems, services)
Coordination / Reservation (policies, intra- subsystem)
Control / Data / Code
NORTH API
SOUTH API
What are CPS research challenges and directions?
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 35
Some literature
• Cyber-Physical Systems: A New Frontier – Lui Sha, Sathish Gopalakrishnan, Xue Liu, and Qixin Wang
2008 IEEE International Conference on Sensor Networks, Ubiquitous,
and Trustworthy Computing
• Cyber Physical Systems: Design Challenges – Edward Lee, 2008 11th IEEE International Symposium on Object Oriented Real-
Time Distributed Computing (ISORC)
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 36
Advances: CPS Systems Science
• A robust design flow
– … combining computational and physical aspects
– … with multi-scale dynamics
– … and integrated wired and wireless networking
• Management of systems’ resources, including e.g. mass, energy
and information
– the system ‘knows’ about itself and about aspects of the real world in
some representation
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 37
Advances: Ubiquitous, trustworthy computing
• Adequate abstractions:
– Distributed, real-time computing
– Real-time group communication
– Dynamic topology management in mobile systems
• Programming models
– that include timing and deal with events in time and space
– have real-time and concurrency abstractions
• e.g. bands of simultaneity
– configurable / tunable software components
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 38
Advances: Robustness, safety and security
• Coherent set of metrics that capture uncertainty, faults / error /
failures and security
– …to be included in the software / system design process
– …and in algorithms (e.g. anytime algorithms, event-based control)
– …to admit optimization, both design and run-time
• A philosophy of a small, correct kernel
– formally specified and verified against a sound and complete
environment model
– against which guarantees for (safety-)critical services are given
• Self monitoring and feedback control within software systems
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 39
Anytime algorithms • The highest quality outcome
of subsystems is not always
required
– (response-)time versus
quality trade-off
• An “any-time” algorithm
– provides a valid solution any
time when
it is interrupted
– increases quality over time
• Examples:
– Newton-Raphson
– Trajectory estimation
– Video decoders
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 40
Advances: composition
• Composition of (software) components includes service quality,
dependencies and resource use
– Rich interfaces, including resource specification, dependencies on
environment conditions
– Resource aware and optimized behavior
• Derive correctness and QoS properties from:
– architecture (logical, physical)
– protocols, mappings
– component properties
• Theory of composition
– language support, model-driven (DSL)
– integrating time-triggered and event-triggered
– systems of systems
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 41
Advances: trust, trustworthiness
• (tools to) Visualize and analyze a CPS within its broader context
(social, technical)
• Transparent privacy protection
– concepts, analytical and engineering framework
• Cohesive, conceptual, predictable, transparent from user’s
perspective
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 42
For all this
better models, better tools
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 43
How does CPS relate to networks?
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 44
Use of data needs networks
1. Improve subsystem operation
– increased knowledge of the system context from external sources
• weather, environment, …
– evaluate effects of actuation and learn to improve
2. Supply detailed, local data to enable new applications (or
improve existing ones), e.g.,
– internal details of production processes
– wear and tear of internal components
– sensing of ‘the real world’
3. Take (increasingly larger) data processing in control loops
– increases in sensor numbers, sampling frequency and sample size
4. (Some) control in the cloud, e.g. cloud based robotics
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 45
Networks CPS • Networks
– form the prime place of integrating subsystems
– are essential ingredients of applications
– are heterogeneous
– connect the ‘physical’ to the ‘Cyber’
• CPS applications are distributed, and there are multiple, e.g.,
– sensors, actuators, control connected by shared networks
– data and management (supervisory control) in the cloud
• Openness, and security for safety and privacy
• Concerns of energy, power, throughput, latency, jitter
– management
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 46
Discussion
15-Sep-16
Johan J. Lukkien, [email protected]
TU/e Informatica, System Architecture and Networking 47