Keywords: Data management, data flows, event management services, data integration

Context: Smartgrids are intelligent energy distribution networks. They are provided with numerous interconnected sensors and computing devices through communication networks that often offer low bandwidth (CLP-3G, CLP-1G: 40 to 200 kb/s, CTN: 50 kb/s, GPRS: 20 kb/s, etc.) These sensors and intelligent meters daily produce large datasets as raw data (e.g. Power voltage or intensity, load curves) or large sets of events (e.g. communication lost, out of limit voltage, open cover). These data/event sets have to be processed with different temporal constraints depending on final uses (e.g. near real time, in one day, over several days). Moreover, these processing may involve all data to give a national view or a local subset based on the network topology (HV-LV Transformation post, Source post, …) or on administrative or internal organization.

The generic functions to be set up concern data acquisition, data storage, data filtering, data correlation, transformation, aggregation depending on specific application requirements such as i) the energy/communication network supervision – which involves of identifying abnormal situations to be reported as events with different priority levels, ii) the supplying various information systems 'connected' to the smart grid, and (iii) the detection of anomalies that may require the integration of various types of data and information.

Objective: All data manipulation (of any kind) have to specified as simply as possible and their processing must keep their efficiency even if the computing environment undergoes important modifications (e.g. CLP network restructuration following a failure in the energy distribution network). A declarative approach for specifying these processes is highly recommended since it offers the best optimization strategies according to the execution environment and not according to the way they are written. E.g. Query evaluation over data should not only produce a tangible result but also information on the way these results have been obtained and on involved data sources (provenance). Therefore the user/application may not only interpret analytical results but also rerun the analysis with different parameters, assumptions and datasets.

These issues cover various scientific fields: sensor networks, event stream composition/aggregation, data definition and manipulation languages, data integration /transformation, high scale data storage, query evaluation over data masses, optimization objective specification/composition, data provenance and traceability, …

The post doc will contribute to the definition and the setup of a distributed data mass management infrastructure in order to facilitate processing specification and execution over continuous data flows coming from a smartgrid. This project is supported by the ERDF  Chaire d’Excellence Industrielle on smartgrids of the Grenoble INP Partnership Fundation (http://www.fondation-grenoble-inp.fr)

Requirements: The candidate must have a PhD and skills in informatics in one or more fields covered by the definition of the future architecture. The candidate should be autonomous since the project consists in the precise specification of the data management requirements for the smartgris. A global knowledge of energy distribution networks and of Smartgrid issues in particular would be appreciated. A training period with ERDF, our industrial partner, to better evaluate these issues, is recommended.

Location: Grenoble Informatics Laboratory (LIG), http://www.liglab.fr

The candidate will have to join the HADAS group of the LIG (specialized in high scale distributed data management).

Duration: 12 months, with a possible 6 months extension.

Starting date: As soon as possible (before September 1^st 2014)

Salary: 3 000 € (not including taxes)

Contact: Christine Collet (Christine.Collet@grenoble-inp.fr)

The application must send by email a detailed CV and a motivation letter.