functional and QoS interoperability at the middleware layer - - PowerPoint PPT Presentation

▶
functional and qos interoperability at the
SMART_READER_LITE
LIVE PREVIEW

functional and QoS interoperability at the middleware layer - - PowerPoint PPT Presentation

Dec 30, 2022 110 likes •468 views

Georgios Bouloukakis In collaboration with: Valrie Issarny and Nikolaos Georgantas Enabling Emergent Mobile Systems in the IoT: functional and QoS interoperability at the middleware layer Emergent mobile systems in the IoT Traffic

slide-1
SLIDE 1

Enabling Emergent Mobile Systems in the IoT: functional and QoS interoperability at the middleware layer

Georgios Bouloukakis

In collaboration with: Valérie Issarny and Nikolaos Georgantas

slide-2
SLIDE 2

2

Emergent mobile systems in the IoT

Emergent Mobile Systems in the IoT – Georgios Bouloukakis

TIM system

Dynamic Heterogeneous

  • Traffic Information Management (TIM) system:
slide-3
SLIDE 3

3

IoT heterogeneity at multiple layers

Emergent Mobile Systems in the IoT – Georgios Bouloukakis

DPWS

Physical layer Data Link layer (MAC & LLC) Network layer Transport Layer Middleware layer Application layer Physical layer Data Link layer (MAC & LLC) Network layer Transport Layer Middleware layer Application layer

IP

slide-4
SLIDE 4

4

Middleware protocols in the mobile IoT

Emergent Mobile Systems in the IoT – Georgios Bouloukakis

DPWS CoAP MQTT ZeroMQ WebSockets …. Client-server Pub/sub Streaming …. reliable/unreliable mobile connectivity ….

slide-5
SLIDE 5

5

Heterogeneous interconnections in the mobile IoT

Emergent Mobile Systems in the IoT – Georgios Bouloukakis

Protocol X Protocol Y

vehicle-devices Yannis

 How to enable interconnections in the mobile IoT ?  What is the end-to-end QoS of the interconnection ?

Bridging middleware protocols Convergence to a single protocol Convergence to a single protocol Relying on a service bus Providing common API abstractions Providing common API abstractions

evaluation of specific protocols and their interconnections performance evaluation in pub/sub systems formal analysis of coupling in distributed architectures

slide-6
SLIDE 6

6

Our solution

Emergent Mobile Systems in the IoT – Georgios Bouloukakis

“Enabling heterogeneous interactions in the mobile IoT calls for automated synthesis of interoperability artifacts as well as evaluation of the interoperability effectiveness in terms of end-to-end QoS”

vehicle-devices

Protocol Y Protocol X

systematic solution to interoperability end-to-end performance analysis

slide-7
SLIDE 7

7

Platform for functional and QoS interoperability

IoT application

  • 1. Automated synthesis of interoperability artifacts

Functional semantics

VSB

Artifacts

  • 2. Formal timed analysis

Timing semantics Formal conditions

  • 3. Performance evaluation

QoS semantics Analytical models Statistical Analysis Simulated models

Emergent Mobile Systems in the IoT – Georgios Bouloukakis

slide-8
SLIDE 8

Automated synthesis of interoperability artifacts

Automated synthesis of interoperability artifacts – Georgios Bouloukakis 8

1

IoT application

  • 1. Automated synthesis of interoperability artifacts

Functional semantics

VSB

Artifacts

  • 2. Formal timed analysis

Timing semantics Formal conditions

  • 3. Performance evaluation

QoS semantics Analytical models Statistical Analysis Simulated models

slide-9
SLIDE 9

Models for core communication styles

Automated synthesis of interoperability artifacts – Georgios Bouloukakis 9

Client–Service (CS)

  • Tight Time & Space Coupling

Publish-Subscribe (PS)

  • Time & Space Decoupling

client server subscriber broker publisher subscriber

Data Streaming (DS)

  • Tight Time & Space Coupling

Tuple Space (TS)

  • Time & Space Decoupling

consumer producer reader tspace writer taker

  • ne-way

two-way sync or async

  • ne-way

two-way stream two-way stream two-way sync

  • ne-way
slide-10
SLIDE 10

Generic Middleware (GM) connector model

Automated synthesis of interoperability artifacts – Georgios Bouloukakis 10

  • Our generic connector defines 4 basic interaction types:
  • ne-way

two-way async two-way sync two-way stream

each interaction is represented as combination of post and get primitives post and get primitives abstract CS, PS, DS and TS primitives

We rely on the GM abstraction to introduce

  • ur middleware protocol interoperability solution
slide-11
SLIDE 11

bus protocol

Our middleware protocol interoperability solution

Automated synthesis of interoperability artifacts – Georgios Bouloukakis 11

  • eVolution Service Bus (VSB)1

vehicle- device PS Protocol X DS Protocol Y

ACCEPT data PUBLISH event

Binding Component 1 Binding Component 2

 BC architecture: relies on GM for automated BC synthesis  Primitives & data conversion between the bus protocol and the Things’ protocols  A universal way to describe the Things’ I/O required

Protocol X Protocol Y

GM connector A GM connector C GM connector B

bus protocol

BC logic BC logic

mobile app

1 G. Bouloukakis et al., ICSOC, 2016

slide-12
SLIDE 12

Automated BC synthesis

Automated synthesis of interoperability artifacts – Georgios Bouloukakis 12

{ "protocol": “Protocol Y”, "operations": { "operation_1": { "type": "stream", "role": “consumer", "scope": "location", "input_data" : " lon,lat" } } }

BC synthesizer

Generic BC

Generic BC logic GM API GM API

GM connector X GM connector Y

GM API

<< Protocol Pool >>

Protocol X Protocol Y Protocol Z

… Concrete BC

Concrete BC logic GM for Bus protocol GM for Protocol Y Bus protocol Protocol Y

  • Generic Interface Description Language (GIDL) & Generic BC
slide-13
SLIDE 13

VSB novelty

Automated synthesis of interoperability artifacts – Georgios Bouloukakis 13

vehicle device MQTT traffic light CoAP estimation service REST Binding Component

MQTT CoAP

Binding Component

CoAP REST

  • Lightweight bus
  • Any bus protocol
  • BCs employed only when necessary
  • Support for any protocol classified under CS, PS, DS & TS
  • Automated BC synthesis
  • 75-96 % person-hours reduction when using VSB
  • Evolution support
  • QoS awareness
slide-14
SLIDE 14

Formal timed analysis of interconnected mobile systems – Georgios Bouloukakis

Formal timed analysis

2

14

IoT application

  • 1. Automated synthesis of interoperability artifacts

Functional semantics

VSB

Artifacts

  • 2. Formal timed analysis

Timing semantics Formal conditions

  • 3. Performance evaluation

QoS semantics Analytical models Statistical Analysis Simulated models

slide-15
SLIDE 15

Timing model for IoT interactions

Formal timed analysis of interconnected mobile systems – Georgios Bouloukakis 15

  • We introduce a unifying timing model for IoT interactions by relying on GM.

 GM one-way timing model:

always connected connection/disconnection limited data lifetime

slide-16
SLIDE 16

GM one-way timing analysis

Formal timed analysis of interconnected mobile systems – Georgios Bouloukakis 16

tpost t’post

lifetime lifetime

δpost tget t’get

time_on

δget

time_off time_on

tget-return

GM sender automaton

post_off post_on delta_post <= max_delta_post post ! delta_post := 0 post_end ? delta_post := lifetime

GM receiver automaton

get_off get_on delta_get <= max_delta_get get ! delta_get := 0 get_end ! delta_get >= time_on delta_get <= time_on delta_get <= time_on get_return ?

slide-17
SLIDE 17

Glue automaton & Verification

Formal timed analysis of interconnected mobile systems – Georgios Bouloukakis 17

  • Sender and Receiver automata interact via the Glue automaton
  • Safety (A[]ϕ) property verified using UPPAAL – necessary condition for failed interactions :

A[] glue.trans_fail imply (sender.post_on and receiver.get_off and delta_post==lifetime and delta_get – time_on>=lifetime)

tpost

lifetime

tget

time_on time_off lifetime

slide-18
SLIDE 18

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 18

Performance evaluation

3

IoT application

  • 1. Automated synthesis of interoperability artifacts

Functional semantics

VSB

Artifacts

  • 2. Formal timed analysis

Timing semantics Formal conditions

  • 3. Performance evaluation

QoS semantics Analytical models Statistical Analysis Simulated models

slide-19
SLIDE 19

19

IoT Interactions across Multiple Layers

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis

  • We enrich our timing model with more realistic constraints found across

multiple layers in the IoT

APP

connection/disconnection

MDW NET

limited data lifetime finite capacity buffers app processing delay finite capacity buffers reliable/unreliable protocols mdw processing delay

  • interop. processing delay

finite capacity buffers transmission delay disconnections always connected connection/disconnection

slide-20
SLIDE 20

Base queueing models for mobile IoT interactions

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 20

𝜇𝑗𝑜

𝜈

𝜇𝑝𝑣𝑢 continuous queue 𝜇𝑝𝑣𝑢 TON / TOFF 𝜇𝑗𝑜

𝜈

intermittent (ON/OFF) queue1,2 𝜇𝑗𝑜

𝜈

𝜇𝑝𝑣𝑢

lifetime expired message valid message

messages exp. queue

  • Additional features:

𝜇𝑗𝑜

𝜈

𝜇𝑝𝑣𝑢

dropped message buffer

finite capacity queue

1 G. Bouloukakis et al., ICC, 2017 2 G. Bouloukakis et al., ICPE, 2017

  • We model the end-to-end path of an IoT interaction by using a combination
  • f different types of queueing models
slide-21
SLIDE 21

DS QoS model for mobile IoT interactions

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 21

  • We model reliable or unreliable interactions by using our queueing models

UNRELIABLE RELIABLE

APP MDW MDW APP

continuous queue

𝐸𝑢𝑠

ON/OFF queue

𝐸𝑞𝑠

continuous queue

𝐸𝑞𝑠

continuous queue

𝐸𝑞𝑠

continuous queue

𝐸𝑢𝑠

ON/OFF queue

𝐸𝑞𝑠

continuous queue

𝐸𝑞𝑠

continuous queue

𝐸𝑞𝑠

If message == expired lifetime

λapp

in

lifetime

λapp

in

If mdw-OFF == true If rec-OFF == true If message == expired If message == expired If message == expired

λapp

  • ut

λapp

  • ut

DS one-way (1W) interactions

slide-22
SLIDE 22

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 22

client server subscriber broker publisher subscriber

consumer producer

reader tspace writer taker

Performance modeling patterns

CS/DS-1w CS-2w async CS-2w sync PS-1w PS-2w stream DS-2w stream TS-1w TS-2w sync

What about heterogeneous interactions?

slide-23
SLIDE 23

One-way PS to DS interconnection

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 23

Bus protocol vehicle- device mobile app PS protocol X DS protocol Y

Binding Component 1 Binding Component 2

PS one-way DS one-way CS one-way

CS-1w unreliable DS-1w reliable PS-1w reliable

slide-24
SLIDE 24

Evaluation Results

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 24

1. ON/OFF queueing model validation 2. One-way PS to DS end-to-end performance evaluation

  • We validate the ON/OFF QM validation through:
  • probability distributions
  • arrival rates extracted from the Orange CDR dataset over Senegal1
  • ON/OFF connectivity traces collected in the metro of Paris2

2 G. Bouloukakis et al., ICPE, 2017 1 G. Bouloukakis et al., WiMob, 2015

slide-25
SLIDE 25

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 25

  • 1. Cité Universitaire → Dugommier; journeys : 34; total duration : 15.18 hours; average

duration journey : 26.8 min; TON = 2.43 min and TOFF = 1.6 min.

  • 2. Dugommier → Cité Universitaire; journeys : 28; total duration : 12.13 hours; average

duration journey : 26 min; TON = 2.5 min and TOFF = 1.2 min.

ON/OFF QM Validation using Connectivity traces (1)

slide-26
SLIDE 26

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 26

  • 2nd path: Dugommier → Cité Universitaire
  • For high rates, there is a quite good match with maximum difference of

about 10%.

ON/OFF QM Validation using Connectivity traces (2)

slide-27
SLIDE 27

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 27

PS to DS performance evaluation: success rates

TON + TOFF = 80 sec

pub pub

TON + TOFF = 30 sec

sub sub

lifetime = 10, 20 and 30 sec λapp = 2 msg/sec

in

  • Success Rate 39%
  • Success Rate 63%
slide-28
SLIDE 28

Performance evaluation of interconnected mobile systems – Georgios Bouloukakis 28

  • Lower lifetime periods produce improved response time (but with lower

success rates)

PS to DS performance evaluation: response times

  • Success Rate 39% and Response Time within 5 sec. with Prob = 0.78
  • Success Rate 63% and Response Time within 5 sec. with Prob = 0.45
slide-29
SLIDE 29

Conclusions & future work

Conclusions & future work – Georgios Bouloukakis 29

4

slide-30
SLIDE 30

Conclusions

30 Conclusions & future work – Georgios Bouloukakis

  • We introduce a platform that enables functional interoperability and QoS-

related interoperability evaluation with focus on the mobile IoT

IoT application

Automated synthesis of interoperability artifacts

Functional semantics

VSB

Artifacts Performance evaluation

QoS semantics Analytical models Statistical Analysis Simulated models

Formal timed analysis

Timing semantics Formal conditions

We enable system designers to:

1. Automatically map functional semantics of heterogeneous Things for integrating them into IoT applications 2. Formally analyze time semantics of heterogeneous IoT interactions for ensuring high success rates 3. Analyze realistic QoS semantics of heterogeneous IoT interactions for assessing end-to-end performance

  • Our platform provides precise design-time modeling, analysis and software

synthesis to ensure accurate runtime system behavior.

slide-31
SLIDE 31

Future Work

31 Conclusions & future work – Georgios Bouloukakis

  • From design for interoperability and design-time evaluation to runtime

adaptation: 1. Dynamic composition of heterogeneous Things in emergency scenarios:

  • face possible emergencies and ensure safety through the composition of Things

2. QoS-aware adaptation of IoT middleware protocols

  • detect performance degradation at runtime and decide appropriate actions

3. Ensure cross-layer resilience for heterogeneous IoT interactions

  • control the underlying IoT networking capabilities to improve and adapt IoT

interactions

4. Explore large-scale IoT deployments

  • explore the deployment of our interoperability, resilience and adaptation

solutions in large-scale IoT applications

slide-32
SLIDE 32

Software artifacts and adoption

32 Conclusions & future work – Georgios Bouloukakis

  • VSB is used as a core component in H2020 CHOReVOLUTION project
  • Download VSB:
  • https://repository.ow2.org/nexus/content/repositories/releases
  • Download Eclipse plugin for defining Things’ GIDLs:
  • http://nexus.disim.univaq.it/content/sites/chorevolution-modeling-notations
  • VSB development and runtime demo:
  • https://youtu.be/UgfM3810RS8
  • Download MobileJINQS:
  • http://xsb.inria.fr/MobileJINQS.jar
  • MetroCognition mobile app:
  • https://play.google.com/apps/testing/edu.sarathi.metroCognition
slide-33
SLIDE 33

Publications (1/2)

33 Conclusions & future work – Georgios Bouloukakis

  • G. Bouloukakis, I. Moscholios, N. Georgantas, V. Issarny, "Performance Modeling of

the Middleware Overlay Infrastructure of Mobile Things", ICC, May 2017, Paris, France

  • G. Bouloukakis, N. Georgantas, A. Kattepur, V. Issarny, "Timeliness Evaluation of

Intermittent Mobile Connectivity over Pub/Sub Systems", ICPE, April 2017, L'Aquila, Italy

  • G. Bouloukakis, N. Georgantas, S. Dutta, V. Issarny, "Integration of Heterogeneous

Services and Things into Choreographies", ICSOC, October 2016, Banff, Alberta, Canada

  • V. Issarny, G. Bouloukakis, N. Georgantas, B. Billet, "Revisiting Service-oriented

Architecture for the IoT: A Middleware Perspective", ICSOC, October 2016, Banff, Alberta, Canada

  • G. Bouloukakis, R. Agarwal, N. Georgantas, A. Pathak, and V. Issarny, "Leveraging

CDR datasets for Context-Rich Performance Modeling of Large-Scale Mobile Pub/Sub Systems", WiMob, October 2015, Abu Dhabi, UAE

slide-34
SLIDE 34

Publications (2/2)

34 Conclusions & future work – Georgios Bouloukakis

  • G. Bouloukakis, R. Agarwal, N. Georgantas, A. Pathak, and V. Issarny, "Towards

Mobile Social Crowd-Sensing for Transport Information Management", NetMob - MIT Media Lab, April 2015, Boston, United States

  • G. Bajaj, G. Bouloukakis, A. Pathak, S. Pushpendra, N. Georgantas, and V. Issarny,

"Toward Enabling Convenient Urban Transit through Mobile Crowdsensing", ITSC, September 2015, Gran Canaria, Spain

  • A. Kattepur, N. Georgantas, G. Bouloukakis, and V. Issarny, "Analysis of Timing

Constraints in Heterogeneous Middleware Interactions", ICSOC, November 2015, Goa, India

  • N. Georgantas, G. Bouloukakis, S. Beauche, V. Issarny, Service-oriented Distributed

Applications in the Future Internet: The Case for Interaction Paradigm Interoperability, ESOCC, September 2013, Malaga, Spain

slide-35
SLIDE 35

Thank you!

MiMove Project Team - https://mimove.inria.fr