• RSSI
• RFID UHF
• Localization RFID
• Smart Shelf
• Smart Drawer

In recent years, RFID systems that are widely applied for the identification of objects and people in radio frequency, are also going to be applied used for localization purposes. In indoor applications (apartments, shopping malls, airports), conventional solutions can use a number of signal parameters: instant of arrival ( Time of Arrival , Time Difference of Arrival, TOA, TDOA), angle information (Angle of Arrival, AOA), phase information (Phase Difference of Arrival , PDOA ) or the amplitude of the received signal (Received Signal Strength Indicator, RSSI). There are also some scenarios with small dimensions where the location can be extremely useful. For example, in a hospital, a better service could be offered through RFID technology, as it can add more control to prevent human errors. Indeed, RFID technology can be useful for correct patient drug supply, dose recording, accurate dispensing, anti-counterfeiting as well as replenishment ordering; besides, it simplifies the information transfer between doctors and nurses (e.g. allergic reactions or drug treatment). In retail industry, real-time inventory based on RFID allows to monitor actual customer demand for products, to prevent an out-of-stock situation by timely replenishing orders, to increase sales through additional services (e.g. fitting rooms with smart mirrors providing suggestions to the customer). In food and restaurant industry, RFID technology allows for a better food control, as for example avoiding expired products sale. In this framework, RFID-based smart shelves, smart freezers, and proximity point readers have been developed in libraries, hospitals and retail industries.
In Chapter I, a brief introduction on RFID systems will be presented, in particular describing the main components involved and the principle of operation. It will be described what is proposed in literature about RFID smart shelf and localization algorithms, with particular attention to the methods exploiting the RSSI information.
In Chapter II, an exhaustive experimental study by using off-the-shelf reader, antennas and tags, will be presented with reference to a wooden drawer filled with drug boxes. The LDA algorithm (supervised classifier) will be compared with the K-Means clustering algorithm (unsupervised classifier), to validate the proposed method. The procedure to get several RSSI average samples during the drawer sliding actions is described, and classification performance is investigated. First of all, an RSSI analysis is described with reference to a static configuration of the drawer (not sliding). Then, two classification algorithms are compared by considering a different number of drawer sub-regions. In the second part, the algorithm exploiting the drawer sliding is described and system performance is illustrated to verify the classification capability in a two-region drawer.
In Chapter III, a localization technique for smart bookshelves based on UHF-RFID systems is presented. Two off-the-shelf reader antennas attached to the bookshelf columns, one in front of the other, are used as an alternative to large-area thin planar antennas integrated onto the shelf top. Two scenarios were considered: the first one with a shelf that is D = 97 cm long, and the second one with D = 150 cm. Exploiting RSSI data acquired by the two antennas, a clustering algorithm is implemented to classify tagged books within one of the regions the shelf has been subdivided into. Preliminary results of the system performance analysis have been compared with simulations to demonstrate that is possible to create an interference region in different sectors on the shelf, through a proper phase shift between the feed currents of the two antennas. The system requires a power divider, a switch, variable phase shifters and finally a fixed or variable power attenuator based on the size of the shelf.
In Chapter IV the algorithm implemented will be described and then the performance results (in terms of normalized confusion matrix) will be presented and discussed. Finally, a preliminary analysis has been presented considering different tags, even if it is still under developing.