12 November 2009. Researchers from a number of European institutions, coordinated by the Distributed Systems Laboratory at the UPM’s Facultad de Informática, are building a services development platform that, once complete, will be capable of processing millions of data per second, as compared with the throughput of tens of thousands of data per second that current technologies support.
This technology could combat real-time credit card fraud, mobile telephony SIM card cloning and even fraudulent unpaid telephone calls, and has many other applications.
Banks and credit card institutions have set up a number of systems to detect fraudulent credit card use. But the systems now in place detect fraud after it has been committed. The policy is to identify the fraudster and contain card user losses. The new system, however, will implement real-time fraud detection, preventing improper credit card use and card holder losses because undue payment will simply not be authorized.
The same applies to mobile telephony. SIM card copying or the fraudulent use of telephone lines is now detected after the fact, with the resulting financial losses. In this case, too, the new technology will manage to stop fraud against a mobile telephony operator and/or user being committed, thereby preventing millions of euros from being defrauded and all the inconvenience to users (submission of claims, prepaid card cancellation, etc.).
Part of the FP7
The real-time services development platform is being developed as part of the European Stream (Scalable Autonomic Streaming Middleware) Project funded by the European Union Seventh Framework Programme. With European funding worth over 3.5 million euros, the project is now at the half-way stage and is scheduled for completion in 2010
The ultimate objective of the project is to build a platform for real-time processing of massive data flows. The key technological innovation is that Stream will be able to use large node clusters to process massive data throughput of the order of millions of data per second. The processing capacity of current single-node technologies is now two orders of magnitude lower than what Stream’s will be.
Apart from coordinating the project, the UPM’s Facultad de Informática laboratory, led by Ricardo Jiménez-Peris, is responsible for developing the scalable data flow processor, which is Stream’s hard core. To do this, it parallelizes the query operators, and can deploy each operator on a 100-node cluster. This multiplies the processable data throughput a hundredfold.
The project is related to cloud computing initiatives. Cloud computing provides computing services over the Internet, enabling users, who are not experts in the management of the resources they use, to access the services available in the Internet cloud. Stream is designed for deployment in a cloud computing environment, with features like elasticity, that is, automatically increasing or decreasing the number of nodes according to the computational requirements at any time, to prevent over-provisioning. Elasticity also helps to cut processing cost to the bare minimum, which is essential in pay-per-use cloud computing.
Other research partners, apart from the Universidad Politécnica de Madrid, are Telefónica I+D, which inputs an antifraud system for mobile telephony and the Greek company Exodus, a subsidiary of the Pireos Bank, which will apply the project results to its credit card payment antifraud systems.
This item on other websites
La Radio am fm 11.11.2009
La Flecha 11.11.2009