Michael has worked for 15 years in the field of Development, Verification and Deployment of 2nd and 3rd generation Radio Access Networks. This work has included development and verification of several flagship features in GSM and WCDMA, which allow fast data throughput on cellular networks. Michael has worked in all parts of the produce life-cycle, from requirements definition to troubleshooting on customer networks. He is currently a verification project manager within LM Ericsson ( Ireland ).

Since the commercial launch of 3G Radio Access Networks in the early part of this decade, the market has been demanding new and more complex features, with ever more demanding time to market requirements. This, combined with demands for continuous improvements in operational excellence, necessitates more efficient development practices. In order to maximise these development efficiencies, it is vital that improvements are also made in verification strategies. 

This presentation will provide a whistle-stop tour along a timeline from 2004 to today, detailing the challenges encountered in verification of new features, and how they were overcome by shifting verification strategies. Lessons learned and problems encountered will also be addressed.  

Five significant shifts in strategy will be covered:

1. Introduction of a generic design strategy and how it changed verification strategy.
2. Move from function based to feature based verification methodology
3. Introduction of parallel testing among multiple organisations and test activities
4. Merge of test organisations within network nodes
5. The future - shifts happening today and just around the corner 

Two case studies will be presented which show how savings in test cost of 70% to 80% were achieved in verification of some features.

Key Points:

Test effectiveness can always be improved, even when testing seems already to be efficient.
Every change in strategy must be evaluated and rolled back if it's shown not to be useful.
Changes in other parts of your organisation may enable introduction of verification efficiencies.

Dr. Valentine Casey is the Global Software Development (GSD) research area leader and a lecturer in Software Quality and Testing at Bournemouth University . He has carried out research in the area of distributed software development for the last10 years. He has published extensively on this topic and his publications include his recent book "Software Testing and Global Industry: Future Paradigms". He previously held a research position with Lero - The Irish Software Engineering Research Centre at the University of Limerick . He has over 20 years experience in the software industry and he has also lectured in the University of Limerick where he received his PhD in Computer Science. He is a Software Engineering Institute trained CMM assessor and holds a MSc. in Software Re-Engineering and a BSc. in Economics and Management from London University .

Today software development has truly become a globally sourced commodity. This trend has been facilitated by the availability of highly skilled software professionals in low cost locations in Eastern Europe, Latin America and the Far East . Organisations endeavouring to leverage the opportunities this provides and to avail of the benefits of establishing operations close to emerging markets have embraced this strategy in large numbers. Software testing plays a key role in delivering high quality products and is a labour intensive, complex and expensive activity. In the context of Global Software Development (GSD) to date testing has been perceived as a well defined task that is relatively straightforward and lends itself to being outsourced or offshored.

This talk considers this specific topic and demonstrates that testing in a GSD environment is not necessarily a straightforward activity. In reality it can be negatively impacted by all the factors associated with distributed software development. In this context practical solutions which can be utilised to address these important issues will be considered and discussed.