Frequency offset tolerant demodulators for UNB communications
Due to the COVID-19 crisis measures the PhD defense of Siavash Safapourhajari will take place (partly) online in the presence of an invited audience.
The PhD defence can be followed by a live stream.
Siavash Safapourhajari is a PhD student in the research group Radio Systems (RS). His supervisor is dr.ir. A.B.J. Kokkeler from the Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS).
Interconnected temperature sensors in a wine cellar, smart meters in houses and wearable sensors monitoring health status are all examples of Wireless Sensor Networks (WSN). Thanks to the progress in wireless communications and networking technologies as well as developments in electronic design, it is possible to deploy numerous low power wirelessly connected devices and sensors for a variety of applications. Nevertheless, improving energy efficiency for power constrained wireless nodes is a never-ending quest. Providing wireless communication for diverse applications with different requirements has led to the emergence of different types of wireless networks. A recently emerged type of wireless networks is Low Power Wide Area Network (LPWAN) which provides a wide coverage (10-50km in rural areas and 1-5 km in Urban areas) and low power communication for low data rate applications.
Different technologies including Ultra-Narrowband (UNB), Spread Spectrum and Narrowband-IoT have been introduced for realizing LPWANs. Among these, UNB communication is one of the interesting candidates as it uses unlicensed spectrum, performs better in presence of interference and provides low cost and wide coverage. The Slow Wireless project focuses on physical layer aspects to investigate UNB communications in the (sub) GHz ISM band. As one of the work packages in the Slow Wireless project, the current thesis aims at digital signal processing techniques for UNB communication nodes. Although the very low data rate in UNB communication relaxes some aspects of the transceiver design (such as lower clock frequencies in the baseband processing), it poses other challenges. One of the prominent challenges in UNB communications is Carrier Frequency Offset (CFO).
CFO might be a consequence of the mismatch between oscillators in the transmitter and the receiver or the Doppler shift caused by the relative movement of the nodes. As a solution to the CFO problem, offset tolerant demodulators have been proposed to replace precise but power-hungry carrier frequency recovery. Most of these methods assume that the CFO is in the same order of the signal bandwidth. Then, instead of precise carrier recovery, an offset tolerant demodulator is used which can tolerate the frequency offset without deteriorating the detection performance. However, in UNB communications using low cost crystals, the resulting CFO can become several times the signal bandwidth. In some UNB systems (e.g. Sigfox) solutions in the network layer are adopted to solve the CFO problem. However, it is not sufficient when low cost crystals are used, and it still places limitations on the communication system design.
To overcome the CFO problem in UNB communication, offset tolerant demodulators are considered in this thesis while focusing on achieving scalable offset tolerance. In other words, the demodulator should obtain the same BER value for a certain Eb/N0 regardless of how large the frequency offset is. Of course, it is assumed that the filter prior to the demodulator is wide enough to capture the signal in presence of a large CFO. In this thesis, first, the performance of two offset tolerant demodulators for FSK and PSK, which are potentially suitable for scalable offset tolerance, are investigated to find their limitations when a large CFO tolerance is needed. For FSK modulation, a DFT-Based demodulator is designed by modifying an existing demodulator. On the other hand, for PSK, an Auto-Correlation Demodulator (ADC) for Double Differential PSK is used which can tolerate CFO. Two aspects are considered for scalable offset tolerance; the BER performance of the demodulator and how the complexity of the demodulator scales when the range of the tolerable offset increases. It is shown that for the DFT-based FSK and the DDPSK demodulator, the complexity and the BER performance are limiting factors, respectively.
To circumvent the limitations of the demodulators for FSK and PSK and to achieve scalable offset tolerance, two different algorithms are proposed. The complexity of the DFT-based demodulator is due to a high complexity window synchronization. Hence, a low complexity window synchronization algorithm is introduced with an efficient implementation. The complexity of the proposed design scales more efficiently compared to the conventional method when tolerable CFO increases. Furthermore, the performance of the DDPSK demodulator degrades as a result of increased noise bandwidth. The increased noise bandwidth is a consequence of increasing the bandwidth of the filter prior to the demodulator which is required in presence of a large frequency offset. To tackle this problem, a demodulator based on shifted correlation is proposed for DDPSK modulated signals. Using this demodulator, the effect of increased noise bandwidth is removed and the BER performance will be independent of the range of the tolerable CFO.
In addition to the CFO, temporal fading in UNB communications must be taken into consideration. Very low data rate communication systems are vulnerable to a time-varying fading channel. Even for nodes at fixed positions in UNB, the movement of surrounding objects can lead to a time-varying fading channel and degrade BER performance. To combat distortion in a time-varying channel, time and frequency diversity techniques together with channel coding and interleaving can be utilized. To afford the redundancy required for these solutions, a higher bitrate is required. Higher order PSK and FSK modulation considerably compromise energy and bandwidth, respectively. Frequency/phase modulation (FPSK) is a method which can increase the bitrate in a more power and bandwidth efficient way than PSK and FSK, respectively.
To alleviate the effect of CFO and temporal fading simultaneously, FPSK modulation is considered in this thesis. As a solution for CFO, an offset tolerant demodulator is proposed for FPSK modulation. Moreover, the performance of this modulation is considered in a time-varying channel while using a system designed for including time diversity. The BER performance of the proposed demodulator is evaluated for different combinations of FSK and PSK modulation orders. Using different scenarios, it is demonstrated that the proposed offset tolerant demodulator for FPSK using time diversity can improve the BER performance in a time-varying channel.
