Trade-off between Efficiency and Robustness Doctoral Colloqium @ - PowerPoint PPT Presentation

Institute of Operating Systems and Computer Networks E ffi ciency E ffi ciency vs Robustness = Trade-o ff Robustness Energy Budget Trade-off between Efficiency and Robustness Doctoral Colloqium @ SenSys’18, Shenzhen Robert Hartung, 2018-11-04

Typical application Robust if ... all packets arrive at the sink within deadline ... all nodes are alive during lifetime of network Sense data Transmit/Receive data Efficient if ... there are few re-transmissions ... there are no collisions Robert Hartung Trade-off between Efficiency and Robustness Seite 2

Typical application Robust if ... all packets arrive at the sink within deadline ... all nodes are alive during lifetime of network Sense data Transmit/Receive data Efficient if ... there are few re-transmissions ... there are no collisions BUT: We can either be robust or efficient Robert Hartung Trade-off between Efficiency and Robustness Seite 2

Challenge 1: Energy constrained Batteries Lifetime estimation (before deployment) State-of-Charge (SoC) estimation (after deployment) Energy Harvesting Reliability / Predictability Robert Hartung Trade-off between Efficiency and Robustness Seite 3

Challenge 2: Environment’s influence Batteries Temperature affects both – Available energy (voltage changes) – SoC estimation Energy Harvesting Depends directly on environment, e.g. sun intensity Node Communication (e.g. losing packets at higher temperatures) Robert Hartung Trade-off between Efficiency and Robustness Seite 4

Challenge 2: Environment’s influence Batteries Temperature affects both – Available energy (voltage changes) – SoC estimation Energy Harvesting Depends directly on environment, e.g. sun intensity Node Communication (e.g. losing packets at higher temperatures) Network performance depends on these challenges! Robert Hartung Trade-off between Efficiency and Robustness Seite 4

REAP: Robust and Effcient networks due to Adaption of Parameters Node Energy Sensors IdealVolting Radio Measurement Transceiver Temperature Pressure Under Volting (temp inv.) Humidity Robert Hartung Trade-off between Efficiency and Robustness Seite 5

REAP: Robust and Effcient networks due to Adaption of Parameters State Energy Reliability RF Channel (Cons., Batt., Harv.) (PRR, Retrans.) (LQI, RSSI) Context (Sensors, Channel, TX Power, ...) Node Energy Sensors IdealVolting Radio Measurement Transceiver Temperature Pressure Under Volting (temp inv.) Humidity Robert Hartung Trade-off between Efficiency and Robustness Seite 5

REAP: Robust and Effcient networks due to Adaption of Parameters State A Energy Reliability RF Channel B C (Cons., Batt., Harv.) (PRR, Retrans.) (LQI, RSSI) Rating D E F Context (Sensors, Channel, TX Power, ...) G Node Energy Sensors IdealVolting Radio Measurement Transceiver Temperature Pressure Under Volting (temp inv.) Humidity Robert Hartung Trade-off between Efficiency and Robustness Seite 5

REAP: Robust and Effcient networks due to Adaption of Parameters Energy Reliability Optimize / Strategy State Strategies A Energy Reliability RF Channel B C Save Energy (Cons., Batt., Harv.) (PRR, Retrans.) (LQI, RSSI) Rating D E Robustness F Context (Sensors, Channel, TX Power, ...) G Tradeoff ... Node Energy Sensors IdealVolting Radio Measurement Transceiver Temperature Pressure Under Volting (temp inv.) Humidity Robert Hartung Trade-off between Efficiency and Robustness Seite 5

REAP: Robust and Effcient networks due to Adaption of Parameters µ DTN Energy Reliability RPL Tasks Scheduler Parameters Network/Transport Layer Optimize / Strategy OS State Strategies A Energy Reliability RF Channel B C Save Energy (Cons., Batt., Harv.) (PRR, Retrans.) (LQI, RSSI) Rating D E Robustness F Context (Sensors, Channel, TX Power, ...) G Tradeoff ... Node Energy Sensors IdealVolting Radio Measurement Transceiver Temperature Pressure Under Volting (temp inv.) Humidity Robert Hartung Trade-off between Efficiency and Robustness Seite 5

Node Under Volting Run below specific operating voltage IdealVolting 1 Under Volting at higher temperatures Saves up to 30% of energy 1 IdealVolting – Reliable Undervolting on Wireless Sensor Nodes, Kulau et. al, ACM Transactions on Sensor Networks (TOSN), 2016 Robert Hartung Trade-off between Efficiency and Robustness Seite 6

State Model Unique to each node Build model on node itself Depends on many parameters Constrained in size and computing power available Additional energy Robert Hartung Trade-off between Efficiency and Robustness Seite 7

Rating Node rates itself in relation to other neighbors Different strategy depending on rating Change duty cycle Use alternative path Rating is be shared with neighbors Robert Hartung Trade-off between Efficiency and Robustness Seite 8

Strategy Send data if deadline is too close (prefer robustness) Reduce sensing frequency if energy is low/critical (prefer efficiency) Delay sending packets to more suitable time slot (e.g. warmer) Robert Hartung Trade-off between Efficiency and Robustness Seite 9

Routing Q A P M E I S J H K C T Robert Hartung Trade-off between Efficiency and Robustness Seite 10

Routing Q A P M E I S J H K C T Robert Hartung Trade-off between Efficiency and Robustness Seite 10

Routing Q A P M E I S J H K C T Develop a RPL OF that choses parent node based on parameters / link rating Robert Hartung Trade-off between Efficiency and Robustness Seite 10

Conclusion Trade-off between efficiency and robustness: Depends on a lot of parameters, e.g. Environmental influences Energy source Neighbors and network topology Application Robert Hartung Trade-off between Efficiency and Robustness Seite 11