-1cmqos provisioning in 0.4cmindustrial wireless sensor ... samuele zoppi | itg fachausschuss 5.2...
TRANSCRIPT
QoS Provisioning inIndustrial Wireless Sensor Networks
Chair of Communication NetworksDepartment of Electrical and Computer EngineeringTechnical University of Munich
Samuele Zoppi, H. Murat Gürsu, Wolfgang Kellerer
Chair of Communication NetworksTechnical University of Munich, Germany
Munich, 1st December 2017
Background
Next-generation industrial automation systems willbe wirelessly interconnected [HPO16].
Industrial NCS.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 2
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Background
Next-generation industrial automation systems willbe wirelessly interconnected [HPO16].
Networked Control Systems (NCS): control loopsclosed over the network.
Industrial NCS.
Controller
Sensor Actuator
Plant
xk uk
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 2
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Background
Next-generation industrial automation systems willbe wirelessly interconnected [HPO16].
Networked Control Systems (NCS): control loopsclosed over the network.
Stochastic LTI control system:
xk+1 = Axk +Buk +wk ,
uk =−Kxk ,
xk plant dynamic, uk control law.
Industrial NCS.
Controller
Sensor Actuator
Plant
xk uk
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 2
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Background
Next-generation industrial automation systems willbe wirelessly interconnected [HPO16].
Networked Control Systems (NCS): control loopsclosed over the network.
Stochastic LTI control system:
xk+1 = Axk +Buk +wk ,
uk =−Kxk ,
xk plant dynamic, uk control law.
Sensor sends xk to the Controller.
Controller computes and sends uk to the Actuator.
Industrial NCS.
Controller
Sensor Actuator
Plant
xk uk
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 2
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Background (2)
Inverted pendulum as benchmark NCS application.
Industrial NCS.
Controller
Sensor Actuator
Plant
xk uk
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 3
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Background (2)
Inverted pendulum as benchmark NCS application.
Sampling frequency: 20 Hz.
Industrial NCS.
Controller
Sensor Actuator
Plant
xk uk
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 3
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Background (2)
Inverted pendulum as benchmark NCS application.
Sampling frequency: 20 Hz.
Uplink traffic:
xk =
xkxkθk
θk
→ 256 bits @ 20Hz = 5 kbps.Industrial NCS.
Controller
Sensor Actuator
Plant
xk uk
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 3
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Background (2)
Inverted pendulum as benchmark NCS application.
Sampling frequency: 20 Hz.
Uplink traffic:
xk =
xkxkθk
θk
→ 256 bits @ 20Hz = 5 kbps.
Downlik traffic:
u→ 64 bits @ 20Hz = 1.25 kbps.
Industrial NCS.
Controller
Sensor Actuator
Plant
xk uk
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 3
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Background (2)
Inverted pendulum as benchmark NCS application.
Sampling frequency: 20 Hz.
Uplink traffic:
xk =
xkxkθk
θk
→ 256 bits @ 20Hz = 5 kbps.
Downlik traffic:
u→ 64 bits @ 20Hz = 1.25 kbps.
WSN (PHY IEEE 802.15.4) link→ 250 kbps.
Industrial NCS.
Controller
Sensor Actuator
Plant
xk uk
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 3
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Motivation
Wireless Sensor Networks (WSN) can support NCS traffic.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 4
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Motivation
Wireless Sensor Networks (WSN) can support NCS traffic.
Control loops pose strict QoS requirements on wireless communications.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 4
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Motivation
Wireless Sensor Networks (WSN) can support NCS traffic.
Control loops pose strict QoS requirements on wireless communications.
WSN suffers from external interference and unreliable links [GVZK16].
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 4
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Motivation
Wireless Sensor Networks (WSN) can support NCS traffic.
Control loops pose strict QoS requirements on wireless communications.
WSN suffers from external interference and unreliable links [GVZK16].
Problem: Current WSN lack dynamic real-time QoS provisioning.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 4
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Motivation
Wireless Sensor Networks (WSN) can support NCS traffic.
Control loops pose strict QoS requirements on wireless communications.
WSN suffers from external interference and unreliable links [GVZK16].
Problem: Current WSN lack dynamic real-time QoS provisioning.
Approach:
1. Definition of a QoS provisioning framework for IWSN.2. Implementation of the framework in a testbed.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 4
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Outline
Background & Motivation
QoS Provisioning Framework
Implementation
Conclusions & Further Work
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 5
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Network Architecture
Centralized, star topology.
GW1 GW2
NMApp1 App2
s1 s2 s3
Network architecture.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 6
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Network Architecture
Centralized, star topology.
Network elements:
1. Application (App): industrial NCS application
2. Network Manager (NM): manager of theNetwork Resources of the entire WSN
3. Gateway (GW): interface btwthe WSN devices, the NM and Apps
4. Sensor (s): WSN device
GW1 GW2
NMApp1 App2
s1 s2 s3
Network architecture.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 6
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Network Architecture
Centralized, star topology.
Network elements:
1. Application (App): industrial NCS application
2. Network Manager (NM): manager of theNetwork Resources of the entire WSN
3. Gateway (GW): interface btwthe WSN devices, the NM and Apps
4. Sensor (s): WSN device
Data links btw NM and WSN devicesthrough the GW.
Control links btw App and WSN devicesthrough the GW.
GW1 GW2
NMApp1 App2
s1 s2 s3
Network architecture.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 6
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (1)
QoSRequirement
App
Scheduler /Radio Res. Manager
NM
LinkQuality Info.
GW
WSNDevice
S
t
f
Radio Resource Model
Control Data
Wireless DetServ
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 7
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (1)
QoSRequirement
App
Scheduler /Radio Res. Manager
NM
LinkQuality Info.
GW
WSNDevice
S
t
f
Radio Resource Model
Control Data
Wireless DetServ
Radio Resource Manager inputs:
1. QoS requirements from the application.2. QoS Model of the MAC radio resources.3. Link Quality Information of the radio resources.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 7
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (1)
QoSRequirement
App
Scheduler /Radio Res. Manager
NM
LinkQuality Info.
GW
WSNDevice
S
t
f
Radio Resource Model
Control Data
Wireless DetServ
Radio Resource Manager inputs:
1. QoS requirements from the application.2. QoS Model of the MAC radio resources.3. Link Quality Information of the radio resources.
Radio Resource Manager outputs:
1. Radio resources for Data packets(application).
2. Radio resources for Control packets(schedules, LQI probes, ...).
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 7
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (2)
Dynamic scheduling is possible in a TDMA-FDMAradio resource grid model. f
t
Beacon Data Probe
Schedules Link Info
Dynamic scheduling protocol.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 8
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (2)
Dynamic scheduling is possible in a TDMA-FDMAradio resource grid model.
Dynamic scheduling protocol:
1. Acquisition of Link Quality Information (input)→ estimated Packet Delivery Ratio→ EWMA for estimation
f
t
Beacon Data Probe
Schedules Link Info
Dynamic scheduling protocol.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 8
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (2)
Dynamic scheduling is possible in a TDMA-FDMAradio resource grid model.
Dynamic scheduling protocol:
1. Acquisition of Link Quality Information (input)→ estimated Packet Delivery Ratio→ EWMA for estimation
2. Acquisition of QoS requirements (input)→ Target application reliability (i.e. 90%)→ Target delay bound (deadline)
f
t
Beacon Data Probe
Schedules Link Info
Dynamic scheduling protocol.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 8
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (2)
Dynamic scheduling is possible in a TDMA-FDMAradio resource grid model.
Dynamic scheduling protocol:
1. Acquisition of Link Quality Information (input)→ estimated Packet Delivery Ratio→ EWMA for estimation
2. Acquisition of QoS requirements (input)→ Target application reliability (i.e. 90%)→ Target delay bound (deadline)
3. Distribution of new schedules (output)→ sequence of radio resources (time-freq. pairs)→ distributed using the beacon→ calculated with a reliability-based scheduler
f
t
Beacon Data Probe
Schedules Link Info
Dynamic scheduling protocol.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 8
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (3) - Scheduling algorithm
Reliability is provided allocating multiple transmissions in the frame.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 9
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (3) - Scheduling algorithm
Reliability is provided allocating multiple transmissions in the frame.
The radio resources are modeled using a scheduling graph:
• Nodes represent time instants before/after time slots.
• Edges represent different frequencies and they are weighted by their PDR.
n0 n1 n2 n3 n4
γ0,0 = 0.1
γ0,1 = 0.7
γ0,2 = 0.1
γ0,3 = 0.6
γ0,4 = γs = 0
︸ ︷︷ ︸1st time slot
γ1,0 = 0.2
γ1,1 = 0.1
γ1,2 = 0.9
γ1,3 = 0.1
γ1,4 = γs = 0
︸ ︷︷ ︸2nd time slot
γ2,0 = 0.2
γ2,1 = 0.1
γ2,2 = 0.1
γ2,3 = 0.8
γ2,4 = γs = 0
︸ ︷︷ ︸3rd time slot
γ3,0 = 0.1
γ3,1 = 0.1
γ3,2 = 0.8
γ3,3 = 0.2
γ3,4 = γs = 0
︸ ︷︷ ︸4th time slot
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 9
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (3) - Scheduling algorithm
Reliability is provided allocating multiple transmissions in the frame.
The radio resources are modeled using a scheduling graph:
• Nodes represent time instants before/after time slots.
• Edges represent different frequencies and they are weighted by their PDR.
A Constrained Shortest Path scheduling algorithm finds the schedule (path)fulfilling the target reliability. → {(0,1) ,(1,3) ,(2,0)}
n0 n1 n2 n3 n4
γ0,0 = 0.1
γ0,1 = 0.7
γ0,2 = 0.1
γ0,3 = 0.6
γ0,4 = γs = 0
︸ ︷︷ ︸1st time slot
γ1,0 = 0.2
γ1,1 = 0.1
γ1,2 = 0.9
γ1,3 = 0.1
γ1,4 = γs = 0
︸ ︷︷ ︸2nd time slot
γ2,0 = 0.2
γ2,1 = 0.1
γ2,2 = 0.1
γ2,3 = 0.8
γ2,4 = γs = 0
︸ ︷︷ ︸3rd time slot
γ3,0 = 0.1
γ3,1 = 0.1
γ3,2 = 0.8
γ3,3 = 0.2
γ3,4 = γs = 0
︸ ︷︷ ︸4th time slot
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 9
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (4) - Results
Simulation results of dynamic scheduling with latency and reliability constraints.
1 2500 5000 7500 10000 12500
Time [no frames]
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Avg
.n
et.
link
succ
ess
pro
bab
ility�
Freq. no : 1,6,11Ptx : -20 dBm
p : 1.0Freq. no : 1,6,11Ptx : -10 dBm
p : 1.0Freq. no : 1,6,11Ptx : 0 dBm
p : 1.0Freq. no : 1,6,11Ptx : 10 dBm
p : 1.0
10 40 70 100 130 160
Delays [ms]
0.0
0.2
0.4
0.6
0.8
1.0
EC
DF
0.900 0.905 0.910 0.915 0.920Application reliability
0.0
0.2
0.4
0.6
0.8
1.0
Network avg.
Delay bound
Network avg.
Target reliability
0
10
20
30
40
50
60
No
sup
por
ted
dev
icesI
22
0.37
Reliability-based scheduling [eaED].
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 10
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (4) - Results
Simulation results of dynamic scheduling with latency and reliability constraints.
WSN operating in a dynamic interference scenario (Wi-Fi APs, @2.4GHz).
1 2500 5000 7500 10000 12500
Time [no frames]
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Avg
.n
et.
link
succ
ess
pro
bab
ility�
Freq. no : 1,6,11Ptx : -20 dBm
p : 1.0Freq. no : 1,6,11Ptx : -10 dBm
p : 1.0Freq. no : 1,6,11Ptx : 0 dBm
p : 1.0Freq. no : 1,6,11Ptx : 10 dBm
p : 1.0
10 40 70 100 130 160
Delays [ms]
0.0
0.2
0.4
0.6
0.8
1.0
EC
DF
0.900 0.905 0.910 0.915 0.920Application reliability
0.0
0.2
0.4
0.6
0.8
1.0
Network avg.
Delay bound
Network avg.
Target reliability
0
10
20
30
40
50
60
No
sup
por
ted
dev
icesI
22
0.37
Reliability-based scheduling [eaED].
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 10
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (4) - Results
Simulation results of dynamic scheduling with latency and reliability constraints.
WSN operating in a dynamic interference scenario (Wi-Fi APs, @2.4GHz).
Dynamic scheduling in presence of increasing Wi-Fi transmission power (Ptx).
1 2500 5000 7500 10000 12500
Time [no frames]
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Avg
.n
et.
link
succ
ess
pro
bab
ility�
Freq. no : 1,6,11Ptx : -20 dBm
p : 1.0Freq. no : 1,6,11Ptx : -10 dBm
p : 1.0Freq. no : 1,6,11Ptx : 0 dBm
p : 1.0Freq. no : 1,6,11Ptx : 10 dBm
p : 1.0
10 40 70 100 130 160
Delays [ms]
0.0
0.2
0.4
0.6
0.8
1.0
EC
DF
0.900 0.905 0.910 0.915 0.920Application reliability
0.0
0.2
0.4
0.6
0.8
1.0
Network avg.
Delay bound
Network avg.
Target reliability
0
10
20
30
40
50
60
No
sup
por
ted
dev
icesI
22
0.37
Reliability-based scheduling [eaED].
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 10
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
QoS Framework (4) - Results
Simulation results of dynamic scheduling with latency and reliability constraints.
WSN operating in a dynamic interference scenario (Wi-Fi APs, @2.4GHz).
Dynamic scheduling in presence of increasing Wi-Fi transmission power (Ptx).
1 2500 5000 7500 10000 12500
Time [no frames]
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Avg
.n
et.
link
succ
ess
pro
bab
ility�
Freq. no : 1,6,11Ptx : -20 dBm
p : 1.0Freq. no : 1,6,11Ptx : -10 dBm
p : 1.0Freq. no : 1,6,11Ptx : 0 dBm
p : 1.0Freq. no : 1,6,11Ptx : 10 dBm
p : 1.0
10 40 70 100 130 160
Delays [ms]
0.0
0.2
0.4
0.6
0.8
1.0
EC
DF
0.900 0.905 0.910 0.915 0.920Application reliability
0.0
0.2
0.4
0.6
0.8
1.0
Network avg.
Delay bound
Network avg.
Target reliability
0
10
20
30
40
50
60
No
sup
por
ted
dev
icesI
22
0.37
Reliability-based scheduling [eaED].
WDetServ guarantees reliability and delay bounds reacting against interference.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 10
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Outline
Background & Motivation
QoS Provisioning Framework
Implementation
Conclusions & Further Work
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 11
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Implementation (1)
Deployment of an WDetServ NCS testbed:
1. Control logic (Controller) in the Cloud.
2. Sensing and Actuation in the WSN devices.
3. Gateway acts as forwarding entity.
4. Inverted Pendulum as benchmark controlapplication.
GW1 GW2
NMApp1 App2
s1 s2 s3
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 12
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Implementation (2)
Problem: several HW and SW latency bottlenecks.
Z1 min OM min Z1 max OM max
Setup
0
5
10
15
Sensorto
Con
trollerDelay
(ms)
L4 / L3
Tx MAC
SPI
Transmission
SPI
Radio
Rx MAC
UART
UART to USB
USB
Cloud USB Cont.
Cloud
Sensor-to-cloud delay measurements[GZO+].
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 13
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Implementation (2)
Problem: several HW and SW latency bottlenecks.
Solution: ad-hoc HW solutions for GW and WSN:• Gateway
high perf., multi-radio, multi-processor−→ low-latency, multi-channel SDR
• Sensorlimited perf., single antenna, single processor−→ Zolertia Z1/RE-Mote, TI SimpleLink
Z1 min OM min Z1 max OM max
Setup
0
5
10
15
Sensorto
Con
trollerDelay
(ms)
L4 / L3
Tx MAC
SPI
Transmission
SPI
Radio
Rx MAC
UART
UART to USB
USB
Cloud USB Cont.
Cloud
Sensor-to-cloud delay measurements[GZO+].
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 13
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Outline
Background & Motivation
QoS Provisioning Framework
Implementation
Conclusions & Further Work
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 14
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Conclusions
NCS traffic can be supported by WSN if QoS provisioning is implemented.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 15
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Conclusions
NCS traffic can be supported by WSN if QoS provisioning is implemented.
Wireless DetServ provides the building blocks for QoS provisioning (latency, reliability, QoC, ...)in WSN.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 15
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Conclusions
NCS traffic can be supported by WSN if QoS provisioning is implemented.
Wireless DetServ provides the building blocks for QoS provisioning (latency, reliability, QoC, ...)in WSN.
The implemented reliability-based scheduler is able to react to changes in the wireless environment.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 15
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Conclusions
NCS traffic can be supported by WSN if QoS provisioning is implemented.
Wireless DetServ provides the building blocks for QoS provisioning (latency, reliability, QoC, ...)in WSN.
The implemented reliability-based scheduler is able to react to changes in the wireless environment.
Latency is the major issue for HW implementation (radio, processing, ext. interface).
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 15
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
Further Work
Measurements of NCS Inverted Pendulum operating over the testbed will be performed.
NCS cross-layer scheduling algorithms will be developed.
Different Link Quality Estimators will be evaluated in the testbed.
Multi-radio, multi-processor, high-speed interface solutions will be implemented.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 16
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich
ReferencesSamuele Zoppi et al. Reliability-based scheduling for delay guarantees in hybrid wired-wirelessindustrial networks. IEEE Transactions on Industrial Informatics, SUMBITTED.
M. Gürsu, M. Vilgelm, S. Zoppi, and W. Kellerer. Reliable co-existence of 802.15.4e TSCH-basedWSN and Wi-Fi in an aircraft cabin. In 2016 IEEE International Conference on CommunicationsWorkshops, pages 663–668, May 2016.
Halit Murat Gürsu, Samuele Zoppi, Hasan Yagiz Ozkan, Yadhunandana R. K., and WolfgangKellerer. Tactile sensor to cloud delay: A hardware and processing perspective. In IEEE ICC 2018SAC Symposium Internet of Things Track (ICC’18 SAC-6 IoT), SUBMITTED.
Mario Hermann, Tobias Pentek, and Boris Otto. Design principles for industrie 4.0 scenarios. InSystem Sciences (HICSS), 2016 49th Hawaii International Conference on, pages 3928–3937.IEEE, 2016.
Samuele Zoppi | ITG Fachausschuss 5.2 workshop on “Cellular Internet of Things” | Munich, Germany 17
Chair of Communication Networks, Department of Electrical and Computer Engineering, Technical University of Munich