Munir Cochinwala

Efficient data reconciliation

Abstract Data reconciliation is the process of matching records across different databases. Data reconciliation requires “joining” on fields that have traditionally been non-key fields. Generally, the operational databases are of sufficient quality for the purposes for which they were initially designed but since the data in the different databases do not have a canonical structure and may have errors, approximate matching algorithms are required. Approximate matching algorithms can have many different parameter settings. The number of parameters will affect the complexity of the algorithm due to the number of comparisons needed to identify matching records across different datasets. For large datasets that are prevalent in data warehouses, the increased complexity may result in impractical solutions. In this paper, we describe an efficient method for data reconciliation. Our main contribution is the incorporation of machine learning and statistical techniques to reduce the complexity of the matching algorithms via identification and elimination of redundant or useless parameters. We have conducted experiments on actual data that demonstrate the validity of our techniques. In our experiments, the techniques reduced complexity by 50% while significantly increasing matching accuracy.

Stream query processing for healthcare bio-sensor applications

The need of a data stream management system (DSMS), with the capability of querying continuous data streams, has been well understood by the database research community. We provide an overview on a DSMS prototype called T2. T2 inherits some of the concepts of an early prototype, Tribeca [M. Sullivan et al. (1998)], developed also at Telcordia, but with complete new design and implementation in Java with an SQL-like query language. Our goal is to build a framework that provides a programming infrastructure as well as useful operators to support stream processing in different applications. We set our first targeted application to healthcare biosensor networks, where we applied T2 to monitoring and analyzing electrocardiogram (ECG) data streams, arriving via wireless networks from mobile subjects wearing ECG sensors. Monitoring remote patients via wireless sensors not only provides convenience and safety assurance to the patients, but also saves health care cost in many aspects.

UCS-Router: a policy engine for enforcing message routing rules in a universal communication system

UCS-Router (E. Bertino et al.) is a policy engine developed for message dispatching in a Universal Communication System. A user of the system has a unique identifier, named Universal Communication Identifier (UCI), independent from the actual communication devices the user owns. It is a task of the UCS-Router to properly route messages or phone calls arriving at the UCI to the proper user devices. The routing process is performed according to policies the user specifies on the type of message, on the device capabilities and states, on the sender characteristics, and on the message sending time. We present a declarative policy language developed for UCS-Router. Users of UCS-Router can use this language for specifying message filtering and routing policies. Moreover, we present issues that can arise in the evaluation process of the policies. Such issues are primarily due to the high expressiveness of the language and to the possibility to handle a great variety of message media format. Finally, we present how the UCS-Router addresses such issues.

Client-Server Caching with Expiration Timestamps

We study client-server caching of data with expiration timestamps. Although motivated by the potential for caching in telecommunication applications, our work extends to the general case of caching data that has known expiration times. Toward this end, we tailor caching algorithms to consider expiration timestamps. Next, we consider several different client-server paradigms that differ in whether and how the server updates client caches. Finally, we perform simulation studies to evaluate the empirical performance of a variety of strategies for managing a single cache independent of the server and for managing caches in a client-server setting.

Modeling cryptographic properties of voice and voice-based entity authentication

Strong and/or multi-factor entity authentication protocols are of crucial importancein building successful identity management architectures. Popular mechanisms to achieve these types of entity authentication are biometrics, and, in particular, voice, for which there are especially interesting business cases in the telecommunication and financial industries, among others. Despite several studies on the suitability of voice within entity authentication protocols, there has been little or no formal analysis of any such methods. In this paper we embark into formal modeling of seemingly cryptographic properties of voice. The goal is to define a formal abstraction for voice, in terms of algorithms with certain properties, that are of both combinatorial and cryptographic type. While we certainly do not expect to achieve the perfect mathematical model for a human phenomenon, we do hope that capturing some properties of voice in a formal model would help towards the design and analysis of voice-based cryptographic protocols, as for entity authentication. In particular, in this model we design and formally analyze two voice-based entity authentication schemes, the first being a voice-based analogue of the conventional password-transmission entity authentication scheme. We also design and analyze, in the recently introduced bounded-retrieval model [4], one voice-and-password-based entity authentication scheme that is additionally secure against intrusions and brute-force attacks, including dictionary attacks.

Near Real–Time Call Detail Record ETL Flows

Telecommunication companies face significant business challenges as they strive to reduce subscriber churn and increase average revenue per user (ARPU) by offering new services and incorporating new functionality into existing services. The increased number of service offerings and available functionality result in an ever growing volume of call detail records (CDRs). For many services (e.g., pre-paid), CDRs need to be processed and analyzed in near real-time for several reasons, including charging, on-line subscriber access to their accounts, and analytics for predicting subscriber usage and preventing fraudulent activity. In this paper, we describe the challenges associated with near real-time extract, transform, and load (ETL) of CDR data warehouse flows for supporting both the operational and business intelligence needs of telecommunication services, and we present our approach to addressing these challenges.

Database performance for next generation telecommunications

Three important trends in telecommunications have received media attention for their potentially wide-ranging impacts. The first is increasing competition among telephone service providers, spurred in the United States by the Telecommunications Act of 1996. The second is the technological convergence between telephony and the Internet. The third is the ever-increasing number of mobile devices and users. In this paper we explore the effects of these trends on telecommunication network databases.

Internet traffic warehouse

We report on a network traffic warehousing project at Telcordia. The warehouse supports a variety of applications that require access to Internet traffic data. The applications include Service Level Agreement (SLA), web traffic analysis, network capacity engineering and planning, and billing. We describe the design of the warehouse and the issues encountered in building the warehouse.

A model-driven approach to rapid service introduction

We present a model-driven approach to rapid service introduction, in which the need for writing and integrating custom integration code in existing operations systems is replaced by interface mapping between service components and management components of operations systems. In particular, services are represented as dependency models, from which mappings to management interfaces of individual OSS components can be derived. Through strategic use of XML, XSLT, and related technologies, our approach is applicable to a wide variety of services and operations systems. A working prototype system has been developed, which includes a custom modeling tool and OSS environment. By separately developing and deploying VoIP and push-to-talk (PTT) services in the prototype, we have successfully demonstrated viability of our approach.

Adaptive resource management of a virtual call center using a peer-to-peer approach

As the number and diversity of end user environments increase, services should be able to dynamically adapt to available resources in a given environment. We present the concepts of migratory services and peer-to-peer connections as the means of facilitating adaptive service and resource management in distributed and heterogeneous environments. Our approach has been realized using object-oriented principles in Adaptive Communicating Applications Platform (ACAP). The architectural design and implementation of a real-life high-level service, Virtual Call Center (VCC), are used to illustrate issues in adaptive service and management issues and discuss in detail our approach in ACAP.