Passive Intermodulation Distortion in Radio Communication Systems: Signal Models and Digital Cancellation

Cellular networks and technology have evolved substantially since the late 1970s, with successive generations bringing a variety of new features. Each upcoming generation is better and more advanced in technology when compared to the previous generation. To this end, a huge amount of research in the field of wireless communications has been dedicated to the following two major aspects: the increasing demand for user data rate requirements and the exponential growth of internet-dependent devices around the world. It is undeniable that the outcome of this phenomenon is a heavy congestion of the available spectral resources. This has inspired the utilization of many innovative solutions for improving the spectral efficiency in wireless communication systems by facilitating simultaneous transmission at a high data rate without having the need for additional spectrum. Some of these technologies include in-phase/quadrature (I/Q) modulation, multiple-input-multiple-output (MIMO) systems, and orthogonal frequency-division multiplexing (OFDM). In part because of these existing solutions, the spectral efficiency of wireless communication has improved considerably. However, more advanced techniques are necessary if future data transfer requirements are to be met, for example in the sub-6 GHz band. Given this, the frequency division duplex (FDD) in Long Term Evolution-Advanced (LTE-Advanced) has suggested carrier aggregation (CA) technology as another step towards the better utilization of the spectral resources. The CA techniques enable the utilization of multiple contiguous or noncontiguous parts of the spectrum, thus allowing for flexibility in transmission bandwidth. By using CA, spectral resources can be combined from the same operating frequency band (intraband CA) or from multiple frequency bands (interband CA). In CA, multiple transmitters and receivers are integrated into a single chip to reduce costs, space, and power consumption. Additionally, the industry is moving towards low-cost radio frequency (RF)- front-end solutions containing duplex filters with limited isolation capabilities in order to cope with ever-increasing CA combinations. This results in the leakage of the own transmit (TX) signal and other so-called blocker signals. One of the challenges in the utilization of the CA is when the combined noncontiguous signal propagates through the TX chain passing through the RF front-end passive components, the cross-modulation of the transmitted signal further creates unwanted passive intermodulation (PIM) products. These PIM products in some cases may fall onto their own receiver (RX) band and may cause receiver desensitization.

In this thesis, we present novel solutions for modeling and suppressing PIM distortion and passive harmonic distortion (PHM) in FDD based radio transceivers. This distortion results from nonlinear RF components and simultaneous transmission and reception, with a particular focus on modern carrier aggregation networks. PIM-induced distortion terms generally have a higher power than the weak received signal, even with state-of-the-art RF components, because of the linearity characteristics of the passive components. Consequently, it is necessary to minimize the harmful impact of such distortion, which can be achieved by taking a variety of approaches. A simple technique can be to decrease the transmit power or relax the receiver reference sensitivity requirements. This is known as maximum power reduction (MPR) and maximum sensitivity degradation (MSD), respectively, for LTE-Advanced and new radio (NR) user equipment (UEs).

In addition to preventing receiver desensitization, these approaches will negatively affect coverage, so they are not very appealing. Uplink (UL) and downlink (DL) resource allocation and scheduling can also be optimized to avoid inband distortion. This approach, however, will be very complex, and could also result in reduced peak throughput and spectrum utilization. The PIM/PHM distortion power can, however, be controlled by improving RF component quality and isolation. Nevertheless, this solution has the drawback of increasing the overall cost of the device. In addition, even with the more expensive components PIM-induced distortion might not be avoided. In this thesis, we take an alternative approach by using the original transmit data as a reference to cancel such PIM in the transceiver digital front-end. For the generation of an accurate cancellation signal, we derive different advanced signal models for the observable intermodulation distortion in the receiver band that incorporate also power amplifier nonlinearities, together with the passive component nonlinearities and the frequency-selective responses of the duplex filters. Real-life RF measurements are conducted with an actual LTE-Advanced user equipment (UE) transceiver system to evaluate the performance and processing complexity of the devised digital cancellation and parameter estimation solutions. The experiments show that PIM-induced self-interference is almost 20 dB above the noise floor even while utilizing state-of-the-art RF components. However, the results show that each of the proposed signal models and the related digital cancellers in this thesis are able to cancel the PIM-induced interference up to 15 dB. The results also indicate that in many cases it is necessary to account for nonlinear distortions caused by amplifiers. This is even if the individual component carriers are combined after the amplification stage. In general, when the nonlinear distortion of the power amplifiers is taken into account the amount of cancellation is further improved by 4-5 dB, in contrast to the linear power amplifier based cancellers.