Packet Priority Assignment for Wireless Control Systems of Multiple - - PowerPoint PPT Presentation

Packet Priority Assignment for Wireless Control Systems of Multiple Physical Systems

Wenchen Wang, Daniel Mosse, and Alessandro V. Papadopoulos

May 8th, 2019 @ISORC, Valencia, Spain

Interconnected World Made of Computing Systems

!2

IoT Connections

4.1 Billion Connections in 2024!!!

Photo-illustration: iStockphoto

404 CAR NOT FOUND

Photo-illustration: iStockphoto

Wired Control System

Plant Sensors Controller Actuators

Targets & Objectives Environment

Wired Control System

Plant Sensors Controller Actuators

Targets & Objectives Environment

Wired Control System

Plant Sensors Controller Actuators

Targets & Objectives Environment

Wired Control System

Plant Sensors Controller Actuators

Targets & Objectives Environment

Wired Control System

Plant Sensors Controller Actuators

Targets & Objectives Environment

Wired Control System

Plant Sensors Controller Actuators

Targets & Objectives Environment

Not easy to deploy and maintain

Wired Control System

Plant Sensors Controller Actuators

Targets & Objectives Environment

At most 4 robots can be connected to 1 cabinet [Salman et al., Fog-IoT 2019] Not easy to deploy and maintain

Wireless Control System (WCS)

Plant(s) Sensors Controller(s) Actuators

Targets & Objectives Environment

Network Network

Delay and Message Loss

0.2 0.4 0.6 0.8 1

time

10 20

PS1

105 0.2 0.4 0.6 0.8 1

time

• 1

1

PS2

0.2 0.4 0.6 0.8 1

time

• 1

1

PS3

Major Challenges of WCS

Instability When the physical system is unstable, the plant or the device can be damaged and leads to serious safety issues and financial loss. Performance Degradation Induced additional error Network-induced error

0.2 0.4 0.6 0.8 1

time

10 20

PS1

105 0.2 0.4 0.6 0.8 1

time

• 1

1

PS2

0.2 0.4 0.6 0.8 1

time

• 1

1

PS3

Major Challenges of WCS

Instability When the physical system is unstable, the plant or the device can be damaged and leads to serious safety issues and financial loss. Performance Degradation Induced additional error Network-induced error

Unstable

0.2 0.4 0.6 0.8 1

time

10 20

PS1

105 0.2 0.4 0.6 0.8 1

time

• 1

1

PS2

0.2 0.4 0.6 0.8 1

time

• 1

1

PS3

Major Challenges of WCS

Instability When the physical system is unstable, the plant or the device can be damaged and leads to serious safety issues and financial loss. Performance Degradation Induced additional error Network-induced error

Unstable Degraded Performance

Problem Formulation

Shared multi-hop network Different paths p_1, p_2, …, p_m Each path with delay D_j TDMA fixed topology Time-varying delivery ratio dr_j

Remote Controller

… Multi-hop Network N Physical Systems (PSs) Dynamic network reconfiguration in [Wang, RTNS 2018]

Different Requirements, Shared Network

time time time Demand Demand Demand LO-Critical HI-Frequency HI-Critical HI-Frequency HI-Critical LO-Frequency

Different Requirements, Shared Network

time time time Demand Demand Demand LO-Critical HI-Frequency HI-Critical HI-Frequency HI-Critical LO-Frequency

Different Requirements, Shared Network

time time time Demand Demand Demand LO-Critical HI-Frequency HI-Critical HI-Frequency HI-Critical LO-Frequency

Different Requirements, Shared Network

time time time Demand Demand Demand LO-Critical HI-Frequency HI-Critical HI-Frequency HI-Critical LO-Frequency

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path

RCA Requested Change Amount

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path

RCA Requested Change Amount RCD Requested Change Duration

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path Objective: Minimize Control performance degradation Induced by the wireless realization Without redesigning the control system

RCA Requested Change Amount RCD Requested Change Duration

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path Objective: Minimize Control performance degradation Induced by the wireless realization Without redesigning the control system

RCA Requested Change Amount RCD Requested Change Duration

RMSEi = 1 Ttrans

Ttrans

∑

t=0

∥yW

i (t) − yWL i

(t)∥2

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path Objective: Minimize Control performance degradation Induced by the wireless realization Without redesigning the control system

RCA Requested Change Amount RCD Requested Change Duration

RMSEi = 1 Ttrans

Ttrans

∑

t=0

∥yW

i (t) − yWL i

(t)∥2

Wired

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path Objective: Minimize Control performance degradation Induced by the wireless realization Without redesigning the control system

RCA Requested Change Amount RCD Requested Change Duration

RMSEi = 1 Ttrans

Ttrans

∑

t=0

∥yW

i (t) − yWL i

(t)∥2

Wired Wireless

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path Objective: Minimize Control performance degradation Induced by the wireless realization Without redesigning the control system

RCA Requested Change Amount RCD Requested Change Duration

RMSEi = 1 Ttrans

Ttrans

∑

t=0

∥yW

i (t) − yWL i

(t)∥2

Wired Wireless Transient

Problem Formulation

The control is more or less difficult based on Setpoint (or reference) tracking Nonlinearity of the controlled system Reliability of the communication path Objective: Minimize Control performance degradation Induced by the wireless realization Without redesigning the control system

RCA Requested Change Amount RCD Requested Change Duration

RMSEi = 1 Ttrans

Ttrans

∑

t=0

∥yW

i (t) − yWL i

(t)∥2

Wired Wireless Transient

∀i ∈ Physical Systems

Solution: Dynamic Packet Priority Assignment

Priority Assignment Static heuristic (baseline) Dynamic heuristic PID Dynamic heuristic Path Selection Network Path Quality Determination

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

Static Heuristic - Baseline

Offline analysis Assuming no packet loss For different requested changes in demand For all PS i compute Assign the priorities that minimise the average Do not change the priorities online

rRMSEi(Tsim) = 1 Tsim

Tsim

∑

j=0

∥ri(j) − yi(j)∥2 rRMSEi(Tsim)

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

Dynamic Heuristic

At every time t, for all PS i compute Sort the PS i by Assign the highest priority to the PS with highest value of

rRMSEi(t) = 1 t

t

∑

j=0

∥ri(j) − yi(j)∥2 rRMSEi(t) rRMSEi(t)

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

PID Dynamic Heuristic

We define the tracking error as The priority for every PS i is computed as strange formula, isn’t it? It is a PID controller!

ei(t) = |ri(t) − yi(t)|

πi(t) = KP (ei(t) + λ t

t

∑

i=1

ei(t)) + KD (ei(t) − ei(t − 1))

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

PID Dynamic Heuristic

We define the tracking error as The priority for every PS i is computed as strange formula, isn’t it? It is a PID controller!

ei(t) = |ri(t) − yi(t)|

πi(t) = KP (ei(t) + λ t

t

∑

i=1

ei(t)) + KD (ei(t) − ei(t − 1))

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

The Path Quality Model: PQModel

After we determine the priority of the measurement packets Includes Network delay Network reliability We compute the path quality for all the paths as

PQ = Dnet + α nloss Δcsp

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

The Path Quality Model: PQModel

After we determine the priority of the measurement packets Includes Network delay Network reliability We compute the path quality for all the paths as

PQ = Dnet + α nloss Δcsp

End-to-end delay

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

The Path Quality Model: PQModel

After we determine the priority of the measurement packets Includes Network delay Network reliability We compute the path quality for all the paths as

PQ = Dnet + α nloss Δcsp

End-to-end delay Control Sampling Period

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

The Path Quality Model: PQModel

After we determine the priority of the measurement packets Includes Network delay Network reliability We compute the path quality for all the paths as

PQ = Dnet + α nloss Δcsp

End-to-end delay Control Sampling Period Consecutive Packet losses

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

The Path Quality Model: PQModel

After we determine the priority of the measurement packets Includes Network delay Network reliability We compute the path quality for all the paths as

PQ = Dnet + α nloss Δcsp

End-to-end delay Control Sampling Period Consecutive Packet losses Design parameter

Remote Controller

… Multi-hop Network

Priority Assignment Path Selection

Case studies

Parameters Inverted Pendula (IP) Nuclear Power Plant (NPP) Sampling period Ts 0.01s 0.1s Simulation time Tsim 100s 300s RCA (6 + 4l) meters (2 + 2l) MWatts RCD 5j seconds 15j seconds ST change [0s, Tsim-RCD] [0s, Tsim-RCD]

l=0,…,4 j=1,…,8 x3 x3

PID Tuning

• 4
• 2

2.6 2.8 3 10-3

RMSE

2 3.2

KD

3.4 5 4 10 6

• 4
• 2

IP

3

KP

4 5 10-3 2 6

KD

7 5 4 10 6 15

• 4
• 2

3

KP

4

• 4

5 10-3 2

• 2

6 7 4 2 4 6 6

• 4
• 2

0.08 0.1

RMSE

2 0.12

KD

0.14 10 4 20 6

• 4
• 2

NPP

0.1

KP

0.15 2 0.2

KD

0.25 10 4 20 6

• 4
• 2

0.1

KP

• 4

0.15 2

• 2

0.2 0.25 4 2 4 6 6

! Tuning

The IP is more sensitive to large delays " decreases with high network interference

Comparison of the Different Heuristics

PID provides better and more stable performance

Improvement With PQModel

The PQ model

• utperforms

end-to-end approaches

Conclusion

We explored the interaction between dynamic packet scheduling and the control system performance in WCS Highly nonlinear systems are not heavily explored in the literature Three heuristics for packet priority assignment PID: most promising Path quality model Tradeoff between delay and reliability

Questions?

Alessandro V. Papadopoulos alessandro.papadopoulos@mdh.se

Remote Controller

…

Multi-hop Network

Priority Assignment Path Selection