SLIDE 1 Guaranteeing Proper-Temporal-Embedding Safety Rules in Wireless CPS: A Hybrid Formal Modeling Approach
Feng Tan*, Yufei Wang*, Qixin Wang*, Lei Bu†, Rong Zheng‡, Neeraj Suri**
* Embedded Systems & Networking Lab, Dept. of Computing, The Hong Kong Polytechnic Univ. † State Key Lab for Novel Software Tech., Dept. of Computer Sci. & Tech., Nanjing Univ., China ‡ Dept. of Computing and Software, McMaster Univ., Canada ** Dept. of Computer Science, TU Darmstadt, Germany
June 26, 2013
SLIDE 2
Evaluation Related Work Background Problem Solution Demand Overview
SLIDE 3
Cyber-Physical Systems (CPS) are typically distributed and life/mission critical. Life/Mission critical CPS demand wireless Wireless is unreliable Conflict
SLIDE 4
Cyber-Physical Systems (CPS) are typically distributed and life/mission critical. Life/Mission critical CPS demand wireless Wireless is unreliable Conflict PTE Safety Guarantee
SLIDE 5
Cyber-Physical Systems (CPS) are typically distributed and life/mission critical. Life/Mission critical CPS demand wireless Wireless is unreliable Conflict PTE Safety Guarantee
Design Pattern Hybrid Modeling
SLIDE 6
Cyber Physical Systems (CPS): systems involving tight/complex coupling of computer and physical subsystems Medical Manufacturing Avionics
SLIDE 7
CPS Features
Typically distributed and life/mission-critical Real-time (in addition to logical time) matters Modeling must integrate both discrete and continuous aspects
SLIDE 8
Distributed life/mission critical CPS demand wireless communications.
SLIDE 9
Distributed life/mission critical CPS demand wireless communications.
SLIDE 10
Distributed life/mission critical CPS demand wireless communications.
SLIDE 11
Distributed life/mission critical CPS demand wireless communications. Wireless is unreliable
SLIDE 12
How to guarantee the safety of life/mission critical wireless CPS? Life/Mission critical CPS demand wireless Wireless is unreliable Conflict
SLIDE 13
How to guarantee the Proper-Temporal-Embedding (PTE) safety rule of life/mission critical wireless CPS?
Life/Mission critical CPS demand wireless Wireless is unreliable Conflict PTE Safety Guarantee
SLIDE 14
What is Proper-Temporal-Embedding (PTE) safety rule?
SLIDE 15
CPS Feature 2: real-time (in addition to logical time) matters!
SLIDE 16
CPS Feature 2: real-time (in addition to logical time) matters!
risky state dwelling time upper bound risky state dwelling time upper bound
SLIDE 17
CPS Feature 2: real-time (in addition to logical time) matters!
enter-risky safeguard interval
SLIDE 18
CPS Feature 2: real-time (in addition to logical time) matters!
exit-risky safeguard interval
SLIDE 19
How to guarantee PTE safety despite of arbitrary wireless link failures?
SLIDE 20
How to guarantee PTE safety despite of arbitrary wireless link failures? Leasing Design Pattern: risky state dwelling time must be leased.
SLIDE 21 General concepts of Leasing design pattern: each CPS entity takes one of the 3 roles. Initiator Supervisor Participant Participant
- 1. request
- 2. lease
- 2. lease
- 3. approve
SLIDE 22
CPS Features: 1. real-time matters; 2. real-time PTE even when aborting/canceling. (+ 3. arbitrary comm. failures)
Initiator Participant Participant
active fallback active fallback active fallback
SLIDE 23
How to formally describe, analyze, and use Leasing design pattern in the context of CPS?
SLIDE 24
How to formally describe, analyze, and use Leasing design pattern in the context of CPS? CPS Feature 3 implies the use of hybrid automata modeling
SLIDE 25
Hybrid Automaton is a state-of-the-art modeling tool for CPS.
Bouncing Ball Example
SLIDE 26
Leasing Design Pattern for PTE Safety Rules: detailed Supervisor's hybrid automaton
SLIDE 27
Leasing Design Pattern for PTE Safety Rules: detailed Initiator's hybrid automaton
SLIDE 28
Leasing Design Pattern for PTE Safety Rules: detailed Participant's hybrid automaton
SLIDE 29
Leasing Design Pattern for PTE Safety Rules: detailed Participant's hybrid automaton
SLIDE 30
Leasing Design Pattern for PTE Safety Rules: detailed Participant's hybrid automaton
SLIDE 31
Validity of the design pattern
Theorem 1: If the temporal parameters of the design pattern hybrid automata satisfy a certain set of linear inequalities, then PTE safety is guaranteed despite of arbitrary communications link failures.
SLIDE 32
Validity of the design pattern
SLIDE 33
Using the design pattern: how to turn design pattern into detailed CPS designs?
SLIDE 34
We proposed a formal procedure to elaborate a design pattern hybrid automaton into a detailed design hybrid automaton.
Elaborate
SLIDE 35
Validity of elaboration
Theorem 2: If detailed design hybrid automata are respectively derived by elaborating corresponding design pattern hybrid automata, then PTE safety is guaranteed despite of arbitrary communications link failures.
SLIDE 36
Laser Tracheotomy Medical CPS: interconnect/interlock smart medical devices to increase safety Laser Tracheotomy without Device Interlock
SLIDE 37
Laser Tracheotomy CPS Laser Tracheotomy Medical CPS: interconnect/interlock smart medical devices to increase safety
SLIDE 38
Demand to use wireless links for safety and efficiency concerns.
Laser Tracheotomy CPS wireless links wireless links
SLIDE 39
Demand to use wireless links for safety and efficiency concerns.
SLIDE 40
Demand to use wireless links for safety and efficiency concerns.
Laser Tracheotomy CPS wireless links wireless links
SLIDE 41
Laser Tracheotomy CPS PTE safety rule.
≥3sec ≥1.5sec ≤60sec
SLIDE 42
System architecture and roles of the design pattern: Initiator, Supervisor, Participant
SLIDE 43
System architecture and roles of the design pattern: Initiator, Supervisor, Participant
SLIDE 44
System architecture and roles of the design pattern: Initiator, Supervisor, Participant
SLIDE 45
System architecture and roles of the design pattern: Initiator, Supervisor, Participant
SLIDE 46
Following the Leasing design pattern and Elaboration procedure, we derive detailed designs
SLIDE 47
Emulation Scheme
SLIDE 48
Emulation Results
SLIDE 49
Related Work
Leasing Protocol [7,8,9,10,11,12][24]
check-point & roll-back logical time vs. real-time PTE uncontrollable physical world parameters
SLIDE 50
Related Work
Use of formal modeling in design pattern [30~33]. Hybrid modeling mostly used for verification [3],[13~16]. Tichakorn [34] proposes use a subclass of hybrid automata for designing periodical hybrid control systems.
SLIDE 51 Conclusion
- 1. Proposed a Lease based design pattern to guarantee
PTE safety rules in wireless CPS, under arbitrary communication link failures.
- 2. Derived the corresponding closed-form linear constraints
for temporal configuration parameters.
- 3. Formal description of design pattern with hybrid modeling.
- 4. Proposed a formal methodology to elaborate design
pattern hybrid automata to detailed design hybrid automata, while maintaining PTE safety properties.
SLIDE 52
Thank you!
Life/Mission critical CPS demand wireless Wireless is unreliable Conflict PTE Safety Guarantee
Design Pattern Hybrid Modeling
SLIDE 53
Cyber Physical Systems (CPS): systems involving tight/complex coupling of computer and physical subsystems Anesthesiology Surgical Medicine Nursing Communications Mechanics Computer Control
SLIDE 54
Cyber Physical Systems (CPS): systems involving tight/complex coupling of computer and physical subsystems Chemical Engineering Control Mechanics Thermal Engineering Communications Computer
SLIDE 55
Cyber Physical Systems (CPS): systems involving tight/complex coupling of computer and physical subsystems Computer Mechanics Aerodynamics Control Material Communications
SLIDE 56
Demand to use wireless links for safety and efficiency concerns.
The Operation Room Spider Web
SLIDE 57
Demand to use wireless links for safety and efficiency concerns.
The Operation Room Spider Web, after medical CPS safety interlocks
SLIDE 58
Demand to use wireless links for safety and efficiency concerns.
Spider Web OR vs. Wireless OR
SLIDE 59
How to guarantee PTE safety despite of arbitrary wireless link failures?
Leasing Design Pattern Hybrid Automata Modeling: formally describe, analyze, and use the design pattern
SLIDE 60
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Supervisor Participant Participant
SLIDE 61
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Supervisor Participant Participant Fallback Fallback Fallback Fallback
SLIDE 62
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Supervisor Participant Participant Fallback Fallback Fallback Request
SLIDE 63
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Supervisor Participant Participant Fallback Request Lease Fallback
SLIDE 64
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Supervisor Participant Participant Fallback Request Lease
SLIDE 65
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Supervisor Participant Participant Request Lease Lease Fallback
SLIDE 66
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Supervisor Participant Participant Request Lease Lease
SLIDE 67
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Supervisor Participant Participant Request Lease Lease Approve
SLIDE 68
General concept of Leasing Design Pattern for CPS PTE guarantee Initiator Participant Participant
active fallback active fallback active fallback
SLIDE 69 The same scenario can also apply to purely cyber
- systems. What's the difference that CPS makes?
Initiator Participant Participant
active fallback active fallback active fallback
SLIDE 70
CPS Features: 1. real-time matters; 2. real-time PTE even when aborting/canceling. (+ 3. arbitrary comm. failures)
Initiator Participant Participant
active fallback active fallback active fallback
SLIDE 71
Leasing Design Pattern for PTE Safety Rules: sketch of Supervisor's hybrid automaton
SLIDE 72
Leasing Design Pattern for PTE Safety Rules: sketch of Initiator's hybrid automaton
SLIDE 73
Leasing Design Pattern for PTE Safety Rules: sketch of Participant's hybrid automaton
SLIDE 74 Emulation Scheme
) ( 5 . 1 ), ( 3 : intervals safeguard PTE ) ( 6 ), ( 35 ), ( 3 : Ventilator ) ( 5 . 1 ), ( 20 ), ( 10 ), ( 5 : Initiator ) ( 3 ), ( 13 : Supervisor
min 1 2 : min 2 1 : 1 , max 1 , max 1 , 2 , max 2 , max 2 , max 2 , max min ,
s T s T s T s T s T s T s T s T s T s T s T
safe risky exit run enter exit run enter req wait fb
SLIDE 75
Example Scenario Patient SpO2 Sensor Ventilator Laser Scalpel Surgeon Supervisor
SLIDE 76
Example Scenario Patient SpO2 Sensor Ventilator Laser Scalpel Surgeon Supervisor
SLIDE 77
Example Scenario Patient SpO2 Sensor Ventilator Laser Scalpel Surgeon Supervisor
SLIDE 78
Example Scenario Patient SpO2 Sensor Laser Scalpel Surgeon Supervisor Ventilator Pausing
SLIDE 79
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
SLIDE 80
Example Scenario Patient SpO2 Sensor Ventilator Pausing
Laser Scalpel Shooting
Surgeon Supervisor
SLIDE 81
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
SLIDE 82
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
lost
SLIDE 83
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
SLIDE 84
Example Scenario Patient SpO2 Sensor Ventilator Laser Scalpel Surgeon Supervisor
SLIDE 85
Example Scenario Patient SpO2 Sensor Ventilator Laser Scalpel Surgeon Supervisor
SLIDE 86
Example Scenario Patient SpO2 Sensor Ventilator Laser Scalpel Surgeon Supervisor
SLIDE 87
Example Scenario Patient SpO2 Sensor Laser Scalpel Surgeon Supervisor Ventilator Pausing
SLIDE 88
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
SLIDE 89
Example Scenario Patient SpO2 Sensor Ventilator Pausing
Laser Scalpel Shooting
Surgeon Supervisor
SLIDE 90
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
SLIDE 91
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
lost
SLIDE 92
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
SLIDE 93
Example Scenario Patient SpO2 Sensor Ventilator Pausing Laser Scalpel Surgeon Supervisor
SLIDE 94
Example Scenario Patient SpO2 Sensor Ventilator Laser Scalpel Surgeon Supervisor