A learning database system to observe malfunctions and to support network planning

Abstract This paper presents a learning database system that can accommodate malfunction observations. Consequently, such observations may be expressed in structured patterns to support network planing which is one of the important network management functions. The underlying system monitors the network protocol tables in order to discover interesting patterns. To achieve this purpose two learning techniques are used. The first technique is empirical and it focuses on data samples by selecting specific fields and subsets of records using structured query language (SQL). Then data abstraction is carried out and interesting characteristics are extracted. The second technique exploits an explanation_based learning (EBL) procedure to obtain operational rules. In this case the domain (network) knowledge is formally expressed and only one training example is analyzed in terms of this knowledge. Thus, the system is capable of discovering various operational patterns, provide sensible advices, and support the network planning activity. Since the monitoring database utilizes a relational model, an integrated computer-aided software engineering (I-CASE) is used throughout the requirement identification, analysis and design phases. Accordingly, the quality of the database system as an engineering product has been achieved. Moreover, the open database connectivity (ODBC) approach is employed in order to provide an efficient interface that allows a client application to access a variety of distributed data sources in addition to its local database.