Low complexity spectrum sensing for cognitive radio enabled aeronautical communication systems

• Main Author: Koodathumkal Mathew Libin
• Other Authors: A S Madhukumar
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/140135
• Institution: Nanyang Technological University

Air traffic has seen tremendous growth over the last decade and is expected to grow continuously. The state of the art of current aeronautical communication system is not capable of handling future developments due to the scarcity of the spectrum resources and demands enhanced air traffic management schemes. The L-band Digital Aeronautical Communication System (LDACS) is gaining traction as a scheme of choice and aims to exploit the capabilities of modern digital communication techniques and computing architectures. To improve the quality of air traffic management, the spectrum required to be used efficiently. Cognitive Radio (CR) based approaches have also been proposed for LDACS to improve spectrum efficiency and communication capacity. The CR-based system enables opportunistic and on-demand access to L-band channel(s) while meeting the reliability and safety requirements. This allows an aircraft to identify vacant spectral bands in the air-to-ground spectrum using spectrum sensing techniques and choose the proper channel to initiate an LDACS air-to-ground transmission, similar to the case with ground-based CRs. However, spectrum sensing for LDACS presents unique challenges compared to terrestrial systems; aircraft in the range of the system is always in motion resulting in a continually varying network structure, while the channel conditions between the communicating entities can change very rapidly, requiring much more complex processing compared to terrestrial CR systems. At the same time, the limited energy budget of the aircraft demands low complexity techniques to improve energy efficiency. Computationally efficient, fast, and reliable detection techniques that perform well in low signal-to-noise ratio and noise uncertainty scenarios are required in LDACS. New spectrum sensing algorithms and their low complexity implementations are proposed in this thesis to address the aforementioned challenges. The first work presents an energy-difference detection-based spectrum sensing scheme for CR-enabled LDACS system. The detection method utilizes the unique shape of the spectrum after removing the legacy system signals from the known spectral gaps. Irregularities of the power spectrum density across continuous narrow bands are detected by comparing the absolute energy difference of the neighbouring LDACS and legacy channels with a predefined decision threshold. Simulation studies show that the proposed energy-difference detection based sensing scheme offers improved detection performance than the conventional Energy Detection (ED) scheme at the low SNR scenarios and identical performance at relatively high SNRs. Though the technique improves the detection accuracy, it suffers from higher computational complexity, also the scheme is applicable only for the sensing of the LDACS spectrum. An adaptive energy detection scheme is proposed as the second work to reduce the computational complexity further, and to provide a generic sensing scheme. In this technique, the historical energy observations from previous sensing epochs are adaptively used to improve the detection accuracy. An enhanced real-time noise variance estimation technique is developed with the aid of cyclic prefix in the LDACS signals. The noise variance can be estimated effectively in real-time irrespective of the primary signal being on or off. The proposed technique does not incur dedicated hardware blocks for noise variance estimation, leading to an efficient hardware implementation of the scheme without significant resource overheads. The proposed technique is integrated into a cognitive radio platform on a field-programmable gate array (FPGA) device to quantify resource overheads. Results of simulation studies show that the scheme provides better detection accuracy compared to the existing ED techniques. To reduce the SNR wall to a lower value compared to the first and second works using the prior knowledge of the preamble signals, a multiplier-less correlator based sensing scheme is proposed. The proposed correlator can also serve as the receiver synchronizer for the LDACS air-to-ground links. The architecture is designed in such a way that it can cater to the preamble structure of LDACS air-to-ground transmissions. The computational precision of the system is enhanced to improve the performance in very low SNR conditions. The proposed architecture offers improved detection performance over traditional energy detection even at low SNR with lower energy consumption than a multiplier-based correlator, while also assisting in receiver synchronization. A simplified cyclostationary detection (CD) scheme is proposed as the fourth work to get the best trade-off between accuracy, complexity, and sensing duration. In this scheme, a sliding correlation-based CD is proposed. The test statistic is a likelihood ratio test (LRT) in which the cyclostationary feature is normalized with the real-time noise variance. The proposed scheme is implemented in Verilog and mapped on a Xilinx FPGA device to quantify the overall resource consumption. Results show that the scheme offers comparable detection performance with the multiplier-less correlator with only one-third of resource overheads and power consumption and better performance than the conventional CD schemes. Wideband spectrum sensing is essential to find the different possible spectral holes in the frequency spectrum of interest. Conventional Nyquist wideband sensing is not a power-efficient solution as it requires massive computing capability due to the higher sampling and quantization and for subsequent processing. In this context, a power-efficient wideband spectrum sensing is proposed to alleviate this issue by providing the best accuracy-complexity trade-off. The current research proposes and implements a wideband sensing architecture based on a one-bit quantization at the CR receiver. A Finite Impulse Response (FIR) filter bank is used to split the wideband to several narrow bands; then, the detection algorithm is applied to each narrow-bands individually. In the proposed scheme, the sampling and quantization unit is reduced to a high-speed comparator, and the multiplication in each filter tap is replaced by a sign changer, which is implemented using a 2’s complement circuit and a 2:1 multiplexer. The scheme considerably reduces the computational complexity of the conventional filter bank based sensing by using a sign changer instead of a multiplier.