Framework for smart surveillance in Konnex environments

This paper aims at the individuation of novel paradigms for Smart Surveillance applications based on collecting intensive sensor dataflows and providing an abstraction layer for event and activity detection. Konnex network will be deployed as the underlying communication and control bus, in order to evaluate its impact on performance and reliability of the system. The objective is to individuate any difficulties that may arise from usage of Konnex networking in surveillance applications and, eventually providing indications and models to overtake the weak points. The surveillance system being investigated here leverages the potentials offered by Radio Frequency Identification (known as RFID) even in home environments, where correct, real-time identification of persons and objects may constitute the basis for an intelligent security system, not limited to intrusion detection, but also monitoring other abnormal activities and events (such as abnormal behaviours of elderly or disabled people, i.e. Alzheimer patients). The paper will show a possible model for surveillance systems moving beyond current commercial products' way of operation. While currently alarm systems merely offer the possibility to switch the full protection on and off, this work proposes a more context aware protection, that tailors itself to the different situations that may take place in a home environment, and a set of smart behaviours to define the best reaction to each event, in order to introduce a novel working model for security systems: an always on operating mode where it's up to the system and not to the user the responsibility to determine whether protection is needed or not. To accomplish this, raw sensor data will undergo a pattern recognition process, where the system gains knowledge even of complex situation, inferring from dataflows coming from the environment (through statistical analysis, artificial intelligence and possibly machine learning). Apart from the always on mode, the introduction of pattern recognition techniques will enable to define a fine grain security model, not only providing intrusion detection, but also an access control policy where the system gains knowledge of which subjects are committing actions on which objects, since each of this two elements can be given an actual value if an identification infrastructure is introduced. The network layer will have to show some specific characteristic in its physical and logical parameters, in order to prove efficient enough to support this security system. Konnex bus will be investigated in this direction, and possibly extended or integrated with the cooperation of other existing network types, both wired and wireless, such as Ethernet/IP, wireless 802.11 and/or 802.15 protocols.