Collision Recovery Receivers for RFIDs

Radio Frequency Identification (RFID) is a very fast emerging technology that wirelessly transmits the identity of a tag attached to an object or a person. It usually operates in a dense tag environment. My work is focused on passive Ultra High Frequency (UHF) tags whose transmission on the Medium Access Control (MAC) layer is scheduled by Framed Slotted Aloha (FSA). In this thesis, I propose the use of multiple antennas at the reader side in order to recover from collision. By exploiting the fact that a tag signal is real-valued while all other components of a received signal are complexed-valued, I have separated real and imaginary part and in that way I have achieved a recovery from a collision that contains a two times higher number of tags than the number of the receive antennas at the reader, under perfect channel knowledge. Furthermore, I have recommended a modification of a tag signal by an additional part that is specially designed to facilitate channel estimation. The recommended method provides excellent results in comparison to perfect channel knowledge. However, due to the constrained set of the designed sequences, a new issue arised. I have investigated the distribution of the additional sequence set within a tag population and depending on that, I have studied different collision scenarios. I have proposed a two phase collision recovery method that takes out all tags with a unique sequence per slot and if there is just one pair of tags with a common sequence left, such collision is resolved by projecting the signal constellation into the orthogonal subspace of the interference. The proposed method improves collision recovery and further increases the system throughput. Moreover, in this thesis I have studied the influence of several parameters on the system throughput, and I have found the maxima of the theoretically expected throughput for receivers with different collision recovery factors and for different receiver architectures. Furthermore, in order to approach to the theoretical maxima, I have proposed spatial filtering in postprocessing. The main intention is to focus separately on different groups of tags by applying different weights within sector postprocessing. In that way tag signals are attenuated or amplified depending on the angle of arrival. I have designed a simple beamformer with the weights modelled by an FFT algorithm and a more complex beamformer with an eigenfilter design. The obtained results show that the reader has become more robust. Additionally, I have derived a semi-analytical formula for calculating the optimal frame size. This formula incorporates properties of the spatial filter and throughput characteristics. In this way, further optimization of frame lengths is achieved. Furthermore,