Formal engineering methodologies for wireless sensor network
development with simulation
Formal engineering methodologies for wireless sensor network
development with simulation
In the current practice in Wireless Sensor Network (WSN) development, the software code representing communication protocols and algorithms tends to be complex, monolithic, and close to low-level operating system and hardware platforms. The software is thus difficult to understand and maintain. This is because of the lack of software engineering (SE) and model-based techniques, tools and infrastructure in the WSN domain. Furthermore, critical design requirements must be guaranteed, as uncertain and unreliable environment circumstances may cause the failure of a WSN deployment, e.g. by node death and communication failures. Therefore, good SE methodologies and techniques including high-level abstraction, separation of concerns, powerful verification and validation (V&V) are needed for WSN development.
This thesis proposes a Formal Co-simulation (FoCoSim-WSN) framework to strengthen current SE practice for WSN development. This framework enables an iterative and incremental development style which combines the benefits of existing simulation and proof-based formal verification approaches and tools. The complexity of software development for the sensor node controller is reduced by separating the controller model from the simulation environment. Controller algorithms for each protocol stack layer can be formally developed and verified in a layered manner using the refinement method of the Event-B language and its RODIN toolkit. The absence of certain classes of faults in controller models which cannot be guaranteed by simulation testing techniques, can be proved by formal methods. The MiXiM simulation of physical environment provides confidence in the reliability and performance analysis through long running simulation via wireless channels.
Our prototype development confirms the flexibility, usability and reusability of the framework for interworking between formal, simulation and co-simulation modelling. Furthermore, the integration of our proposed FoCoSim-WSN framework with the Model-Based Trace Testing (MBTT) approach gives us confidence in the validation coverage. Test scenarios including functional, failing and recovery tests are created from the sequence of events in our co-simulation master algorithm. Long-running test scenarios generated by MiXiM co-simulation enable model debugging for absent or erroneous constraints and events in our formal controller. Finally, by investigating two case studies we identify reuse opportunities and propose reusable patterns for Event-B and master models.
Intana, Adisak
2fe6c75f-74c5-4d5d-a97e-adc1623654bb
December 2015
Intana, Adisak
2fe6c75f-74c5-4d5d-a97e-adc1623654bb
Poppleton, Michael
4c60e63f-188c-4636-98b9-de8a42789b1b
Intana, Adisak
(2015)
Formal engineering methodologies for wireless sensor network
development with simulation.
University of Southampton, Physical Sciences and Engineering, Doctoral Thesis, 535pp.
Record type:
Thesis
(Doctoral)
Abstract
In the current practice in Wireless Sensor Network (WSN) development, the software code representing communication protocols and algorithms tends to be complex, monolithic, and close to low-level operating system and hardware platforms. The software is thus difficult to understand and maintain. This is because of the lack of software engineering (SE) and model-based techniques, tools and infrastructure in the WSN domain. Furthermore, critical design requirements must be guaranteed, as uncertain and unreliable environment circumstances may cause the failure of a WSN deployment, e.g. by node death and communication failures. Therefore, good SE methodologies and techniques including high-level abstraction, separation of concerns, powerful verification and validation (V&V) are needed for WSN development.
This thesis proposes a Formal Co-simulation (FoCoSim-WSN) framework to strengthen current SE practice for WSN development. This framework enables an iterative and incremental development style which combines the benefits of existing simulation and proof-based formal verification approaches and tools. The complexity of software development for the sensor node controller is reduced by separating the controller model from the simulation environment. Controller algorithms for each protocol stack layer can be formally developed and verified in a layered manner using the refinement method of the Event-B language and its RODIN toolkit. The absence of certain classes of faults in controller models which cannot be guaranteed by simulation testing techniques, can be proved by formal methods. The MiXiM simulation of physical environment provides confidence in the reliability and performance analysis through long running simulation via wireless channels.
Our prototype development confirms the flexibility, usability and reusability of the framework for interworking between formal, simulation and co-simulation modelling. Furthermore, the integration of our proposed FoCoSim-WSN framework with the Model-Based Trace Testing (MBTT) approach gives us confidence in the validation coverage. Test scenarios including functional, failing and recovery tests are created from the sequence of events in our co-simulation master algorithm. Long-running test scenarios generated by MiXiM co-simulation enable model debugging for absent or erroneous constraints and events in our formal controller. Finally, by investigating two case studies we identify reuse opportunities and propose reusable patterns for Event-B and master models.
Text
phdThesis_Adisak.pdf
- Other
More information
Published date: December 2015
Organisations:
University of Southampton, Electronic & Software Systems
Identifiers
Local EPrints ID: 387248
URI: http://eprints.soton.ac.uk/id/eprint/387248
PURE UUID: 1a1a938c-30ca-4fbc-afb7-d626f0d188de
Catalogue record
Date deposited: 18 Feb 2016 12:41
Last modified: 14 Mar 2024 22:46
Export record
Contributors
Author:
Adisak Intana
Thesis advisor:
Michael Poppleton
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics