User generated big data is essential for cheaper security solutions
- 31 July, 2016
- TU Delft Amsterdam Security
Blog of Prof.dr. Pieter van Gelder, director of TU Delft Safety and Security Institute
We are looking for cheaper security solutions with higher efficiency, and my personal opinion is that user generated big data is essential for achieving this.
The first attempt to model big data was by the Frenchman André-Michel Guerry already in 1833 in his study ‘Crimes contres les propriétés, from Essai sur la statistique morale de la France’, who was particularly interested in uncovering the relation between social and moral variables. How are personal crime and property related to each other, donations to the poor, wealth, and so forth? Although this was not really big data, it was a big data study avant la lettre.
Nowadays, we go from a descriptive analysis (based on reporting), to real time monitoring (camera’s, the citizen as interactive sensor) to early warning and predictive analysis (extrapolation and self learning systems), in which very big data, exa bytes (which is a megabyte of a megabyte of a megabyte) even up to zetta bytes are being processed. Figure 1 shows the abundance of incoming observations at the Real Time Crime Center in Manhattan, USA.
Large-scale deployment of autonomous technology, sensors and robotics can gather data by a vast array of sensors including omnidirectional cameras, optical character recognition, thermal imaging, air quality, machines can collect significant quantities of real world data providing organizations with historic and real time information, behavioral analysis, and user-defined alerts enabling improvements in intelligence and analytics, and ultimately serving to make better decisions.
Not only land-based machines, but also unmanned aerial vehicles or drones, the very small ones (Fig. 2) being developed at the faculty of Aerospace Engineering of TU Delft, can be used to improve safety and security, by collecting big data.
Multifunctional radars are also examples of the big data trend. The MIMO-SAR based radar (developed by prof. Yarovoy’s group at Electrical Engineering, TU Delft), added with neural network based detection systems can be used to detect concealed weapons, for instance at airports. The undesired electromagnetic fields of these types of radars is a factor 100 less than the field of a mobile telephone and they provide 3D scans of the human body, even when the person is in motion.
Ultra wideband radars are developed for human being detection, based on cardio or breathing spectrum analysis, very useful during recovery work after earthquakes or tornado’s for instance, but also to detect human trafficking.
The above shows that mass data collection may be very helpful for safety and security applications, but we realize that it may scare people because of privacy violations and the cross correlations which can be made to other datasets to further identify details of people. If we look at camera surveillance in city centres, the standard way of detecting faces is by identification with so-called eigenfaces (based on Principal Component Analysis). Privacy preserving signal processing is possible which allows for the detection of criminals, without leaving the faces of innocent people in the data base. These kind of cryptographic algorithms should be further developed in order to minimize possible negative consequences after a successful hack, as well as better cyber security protections to reduce the probability of a breach as much as possible.
I look into the future with many interesting challenges ahead. I am convinced that further technological developments both on the preventive and repressive sides of security breaches, intelligent monitoring and a more adequate system of systems approach (connecting human factors, technology, organization, ethics, economy and law) will improve the safety and security of our society further in the decades to come.
Noot voor de redactie / niet voor publicatie:
Voor meer informatie kunt u contact opnemen met:
Alice Muurlink, Communication Manager