Return to search

Self-interference channel and analog baseband cancellation for full duplex transceiver

Full duplex (FD) radios are the next generation wireless paradigm to answer the growing demand of high capacity along with energy and spectrum efficient wireless transceivers. Given the colossal power difference between the transmit and receive signal, self-interference cancellation becomes one of the key challenge in the design of a FD radio. A model of self-interference channel is required to develop a robust cancellation mechanism. One of the key contribution of this thesis is to define the properties of the self-interference channel. Furthermore, an analog baseband cancellation mechanism for FD transceivers is also defined, which can be used as a cancellation stage before the signal goes to digital domain.

The self-interference channel was measured using ultra wide-band antennas (UWB). Narrow-band measurement technique i.e., a vector network analyzer (VNA) was used for the channel measurements. Spatial resolution of 4.3 cm was achieved. Measurements were done in variety of locations including an anechoic chamber with different antenna orientation. Antennas were mounted on an old laptop frame. Coherence bandwidth of the self-interference channel was found to be varying between 1 MHz and 10 MHz, effectively making it a frequency selective channel. The average amount of isolation was found to be around 40 dB irrespective of the antennas relative orientation. It was also observed that a major amount of power was transferred because of direct coupling between the antennas and this coupling was due of the frame on which antennas were mounted.

Using the defined analog baseband cancellation mechanism, an orthogonal frequency division multiplexing (OFDM) based transceiver was simulated using Matlab. The impact of different bit analog-to-digital converter (ADC), digital-to-analog converter (DAC), different training sequence length for the desired and the self-interference channel, were observed in the simulations. The simulations were performed for both 16 and 64 quadrature amplitude modulation (QAM). Training symbols were used in front of the data frame to estimate both the desired and self-interference channel and also to set the gain of variable gain amplifiers (VGA). The least square algorithm was used for the estimates. The self-interference power was set to −20 dBm and thermal noise floor was set to −81.68 dBm. It was found that a twelve bit ADC along with a sixteen bit DAC would provide a performance within 1.5 dB of theoretical performance.

Identiferoai:union.ndltd.org:oulo.fi/oai:oulu.fi:nbnfioulu-201308311677
Date02 September 2013
CreatorsSethi, A. (Alok)
PublisherUniversity of Oulu
Source SetsUniversity of Oulu
LanguageEnglish
Detected LanguageEnglish
Typeinfo:eu-repo/semantics/masterThesis, info:eu-repo/semantics/publishedVersion
Formatapplication/pdf
Rightsinfo:eu-repo/semantics/openAccess, © Alok Sethi, 2013

Page generated in 0.0017 seconds