Recorp: Receiver-Oriented Policies for Industrial Wireless Networks - - PowerPoint PPT Presentation

▶
recorp receiver oriented policies for industrial wireless
SMART_READER_LITE
LIVE PREVIEW

Recorp: Receiver-Oriented Policies for Industrial Wireless Networks - - PowerPoint PPT Presentation

Dec 22, 2023 308 likes •462 views

Recorp: Receiver-Oriented Policies for Industrial Wireless Networks Ryan Brummet*, Octav Chipara, Ted Herman University of Iowa | Mobile Systems Laboratory Industrial Wireless Networks Applications process control systems Workload

slide-1
SLIDE 1

University of Iowa | Mobile Systems Laboratory

Recorp: Receiver-Oriented Policies for Industrial Wireless Networks

Ryan Brummet*, Octav Chipara, Ted Herman

slide-2
SLIDE 2

University of Iowa | Mobile Systems Laboratory

  • Applications

– process control systems

  • Workload

– stable periodic flows – known period, deadline, and phase

  • Strict performance requirements

– predictability – high reliability – real-time

Industrial Wireless Networks

2

plantengineering.com

slide-3
SLIDE 3

University of Iowa | Mobile Systems Laboratory

  • Moving machinery

Challenges: Network Dynamics

3

  • Outdoor environments

Tesla Automation Automation.com

  • Moving workers

Ford Motor Company

slide-4
SLIDE 4

University of Iowa | Mobile Systems Laboratory

  • Time Slotted Channel Hopping

– time division multiple access with channel hopping – predictable – centralized

  • Limited Flexibility

– overprovisioned retransmissions to handle link variability

TSCH Schedules

4

F1: CB CH 2 CH 1 F0: AB F1: CB F0: AB CH 0 3 2 1 Slot

B A C

slide-5
SLIDE 5

University of Iowa | Mobile Systems Laboratory

TSCH Schedules

5

  • State of the art

– sacrifice predictability for flexibility – examples – (1) slot stealing – (2) hybrid TDMA with CSMA – (3) low likelihood transmissions in the same slot

  • Can we do better?
slide-6
SLIDE 6

University of Iowa | Mobile Systems Laboratory

Allow transmissions to be reallocated conditioned

  • n the local state at runtime

Key Insight

6

slide-7
SLIDE 7

University of Iowa | Mobile Systems Laboratory

Key Insight

7

F0: AB F1: CB CH 2 CH 1 F0: AB F1: CB F1: CB F0: AB CH 0 3 2 1 Slot

B A C

  • Coordinator initiates transmissions

– coordinator pulls for packets – packet transmitted upon pull reception

  • Transmissions selected via local state

– selections prioritized via a priority

  • rdered service list of transmissions
  • Offline synthesis

– coordinators and service lists ensure reliable packet delivery

slide-8
SLIDE 8

University of Iowa | Mobile Systems Laboratory 8

Run-Time Adaptation

  • Schedule

– packets dropped in two traces

F0: AB

——

F0: AB F1: CB F1: CB F0: AB F0: AB Spec F1: CB F1: CB F0: AB

Trace 3 Trace 2

F1: CB

—— ——

F1: CB

——

F0: AB

Trace 1

3 2 1 Slot F0: AB F1: CB F0: AB F1: CB F0: AB F1: CB F0: AB F1: CB F0: AB Spec

—— ——

F0: AB

Trace 3 Trace 2

F0: AB F1: CB

—— ——

F1: CB F0: AB

Trace 1

3 2 1 Slot

  • Recorp policy

– no packets dropped

slide-9
SLIDE 9

University of Iowa | Mobile Systems Laboratory

Recorp Design

9

Policy Builder Policy Evaluator

Workload

Current state Pull operations Policy

Offline synthesis

(§3.B.1) (§3.B.2)

Runtime adaptation (§3.A) Unschedulable

Routes

  • Min. Link

Quality

slide-10
SLIDE 10

University of Iowa | Mobile Systems Laboratory

  • Simulation

– 41 nodes, 1 base station – 50 flows – 3 flow periods – 3 different workload scenarios – 100 runs – ensure 99% end-to-end reliability

  • Workload Scenarios

– collection (COL) – dissemination (DIS) – route through the base station (RTB)

Evaluation

10

  • Comparison

– schedules – flow centric policies (FCP)

  • Measurement

– real-time capacity – relative to schedule performance

slide-11
SLIDE 11

University of Iowa | Mobile Systems Laboratory

Real-Time Capacity

11

slide-12
SLIDE 12

University of Iowa | Mobile Systems Laboratory

  • Recorp policies

– distribute retransmissions at run-time in response to network dynamics – utilize local adaptation to distribute allocated transmissions

  • Significant performance improvement compared to state-of-the-art

– 1.63 to 2.44 times median increase in real-time capacity

  • For more details please see our paper

Conclusion

12

slide-13
SLIDE 13

University of Iowa | Mobile Systems Laboratory

Worst-Case Response Time

13

slide-14
SLIDE 14

University of Iowa | Mobile Systems Laboratory

State Example

14

F0: F F1: F F0: F F1: F F0: S F1: F f F0: F F1: F f F0: S F1: F s s s PLA(F0,F1) F0: S F1: S f F0: F F1: F F0: S F1: F f s f s PLA(F0,F1) F0: S F1: S 1 F0: F F1: S F0: F F1: F f s 1 f s

1 Pf Pf 2 Pf 3 3PsPf 3 +Pf 4 Ps Ps 2 Ps 2 +2Ps 2Pf 2PsPf 3PsPf 2 Ps 2 +2Ps 2Pf+3Ps 2Pf 2 +PsPf 3

PLA(F1) S3 Slot 0 Slot 1 Slot 2 Slot 3 PLA(F0) S2 S2 S1 S0 S0 S1 S1 S0 S0 S0 Release F0, F1 Complete F0 Complete F1