SD-CPS: Taming the Challenges of Cyber-Physical Systems with a Software-Defined Approach

SD-CPS: Taming the Challenges ofCyber-Physical Systems

with a Software-Defined Approach

Pradeeban Kathiravelu1,2, Luıs Veiga1

1Instituto Superior Tecnico, Universidade de Lisboa, Portugal2Universite catholique de Louvain, Louvain-la-Neuve, Belgium

4th IEEE International Conference on Software Defined Systems (SDS 2017).May 8 - 11, 2017. Valencia, Spain.

Pradeeban Kathiravelu (IST-ULisboa) SD-CPS 1 / 24

Introduction

Cyber-Physical System (CPS)

A system composed of cyber and physical/mechanical elements.Internet of Things (IoT).Industry 4.0.

1

1Christoph Roser at AllAboutLean.comPradeeban Kathiravelu (IST-ULisboa) SD-CPS 2 / 24

Introduction

Challenges in CPS

Modelling and designing the systems.Management and orchestration of the intelligent agents.

Large-scale heterogeneous execution environments.Unpredictable.Decision making: communication and coordination.

Security and distributed fault-tolerance.Recovery upon system and network failures.


Motivation

Motivation

A novel approach to address the challenges of CPS.


Motivation

Motivation

A novel approach to address the challenges of CPS.

Desired Properties in a new CPS ApproachEasy to adopt from current CPS approaches.Should not introduce more/new challenges.


Motivation

Our Proposal!

Software-Defined Cyber-Physical Systems (SD-CPS)


Motivation

Our Proposal!

Software-Defined Cyber-Physical Systems (SD-CPS)Why it is?

How it is?


Motivation

Our Proposal!

Software-Defined Cyber-Physical Systems (SD-CPS)An architectural enhancement to tackle the corechallenges of CPS.

Bring SDN to CPS!A middleware framework for CPS.Inspired by the centralized control offered by SDN.


Motivation

Contributions of SD-CPS

A dual (physical and virtual/cyber) execution environment.An efficient approach to CPS modelling and simulations.Mitigate the unpredictability of the physical execution environment.

Enhanced communication and coordination of autonomous agents.Virtually separated control and data flows.A clear logical separation of devices from their execution.

Resilience for critical flows with a differentiated QoS.End-to-end delivery guarantees.


Architecture

SD-CPS Controller Architecture


Architecture

SD-CPS

Taming the Challenges of CPS with SD-CPS!


Architecture

1. Controlling the CPS Data Plane


Architecture

2. Scaling the CPS with SD-CPS

Controller Farm: A federated deployment of SDN controllers.SDN + Message-Oriented Middleware.Protected access among controller clusters in wide area.Scaling the control plane - Flat; without a hierarchy.

Software-Defined Sensor Networks.Extend SDN for sensor networks.SDN with sensors and actuators.


Architecture

3. Modelling and Simulating CPS

Cyberspace to model the smart devices as virtual intelligent agents.Interactions between the actors in physical and cyber spaces mapped.Incrementally model and load from the controller farm.


Architecture

3. Modelling and Simulating CPS

Orchestrator: Build once, deploy and execute twice (or more).Software-Defined Simulations ⇒

Software-Defined Service Compositions.Tenant-Aware Virtual Network Allocation.

Multiple isolated virtual execution environments.Pradeeban Kathiravelu (IST-ULisboa) SD-CPS 15 / 24

Architecture

4. Resilience in SD-CPS


Architecture

4. Resilience in SD-CPSCase 1: Sending the subflow to the destination node


Architecture

4. Resilience in SD-CPSCase 2: Sending the subflow to an intermediate node


Architecture

5. Security in SD-CPS

Distributed fault tolerance and recovery upon system failures.Differentiated bandwidth provisioning with QoS awareness.

Global knowledge of CPS to the controller.Controller as the most valuable entity in the CPS.Avoid introducing vulnerabilities.

Security researches on SDN and CPS are applicable.


Implementation

Prototype Implementation and Feasibility Assessment

Assessments with simulations and microbenchmarks.Increased controller performance.

CHIEF [SDS’16] as the controller farm.based on OpenDaylight, AMQP, . . ..

Increased orchestration/centralized management capabilities.Cassowary [M4IoT’15] for Software-Defined Sensor Networks.

Reduced Modelling Time.Dual reality of cyber-physical spaces for simulations and designs.

SDNSim [CoopIS’16] for Software-Defined Simulations.SENDIM [IC2E’16] to offer the orchestration capabilities.

Performance growth = f(problem size, workflow as services).Mayan [ICWS’16] to offer Software-Defined Service Compositions.


Conclusion

Related Work

SDN for Heterogeneous Devices.Sensor OpenFlow (Software-Defined Wireless Sensor Networks).

SDN + WSN = SD-WSN.Scaling SDN.

Federated Controller DeploymentsClustering SDN controller with Akka.OpenDaylight Federation and Conceptual Data Tree projects.

Software-Defined Systems for Smart Environments.Software-Defined Buildings (SDB).Software-Defined Environments (SDE).Software-Defined IoT

SDSecurity + SDStorage + SDN + . . . = SDIoT.Taming the challenges of distributed systems.

Albatross.


Conclusion

Conclusion and Future Work

Summary

An approach to mitigate the design and operations challenges in CPS.Benefits of SDN to CPS.

Unified and centralized control.Improved QoS, management, and resilience.

Reduced repeated effort in modelling.


Conclusion


Summary




Future WorkIncreased resource efficiency using locality data in the Internet scale.Metering and billing the tenant resource consumption in CPS.Securing SD-CPS in the presence of malicious participants.


Conclusion


Summary




Future WorkIncreased resource efficiency using locality data in the Internet scale.Metering and billing the tenant resource consumption in CPS.Securing SD-CPS in the presence of malicious participants.

Thank you! Questions?


Engineering

SD-CPS: Taming the Challenges of Cyber-Physical Systems with a Software-Defined Approach