Design and implementation of components for renewably-powered base-stations with heterogeneous access channel

  • Vianney ANIS

Student thesis: Doctoral Thesis

Abstract

Providing high-speed broadband services in remote areas can be a challenging task, especially because of the lack of network infrastructure. As typical broadband technologies are often expensive to deploy, they require large investment from the local authorities. Previous studies have shown that a viable alternative is to use wireless base stations with high-throughput point to point (PTP) backhaul links. With base stations comes the problem of powering their systems, it is tackled in this thesis by relying on renewable energy harvesting, such as solar panels or wind turbines. This thesis, in the context of the sustainable cellular network harvesting ambient energy (SCAVENGE) project, aims to contribute to a reliable and energy efficient solution to this problem, by adjusting the design of an existing multi-radio energy harvesting base station. In Western Europe, 49 channels of 8 MHz were used for analogue TV transmissions, ranging from 470 MHz (Channel 21) to 862 MHz (Channel 69); this spectrum, now partially unused due to the digital television (DTV) switch-over, has been opened to alternative uses by the regulatory authorities. Using this newly freed ultra high frequency (UHF) range, also known as TV white space (TVWS), can offer reliable low-cost broadband access to housings and businesses in low-density areas. While UHF transmitters allow long range links, the overcrowding of the TV spectrum limits the achievable throughput; to increase the capacity of such TVWS rural broadband base station the UHF radio has previously been combined with a lower-range higher throughput GHz radio like Wireless Fidelity (WiFi). From the regulatory constraints of TVWS applications arises the need for frequency agile transceivers that observe strict spectral mask requirements, this guided previous works towards discrete Fourier transform (DFT) modulated filter-bank multicarrier (FBMC) systems. These systems are numerically efficient, as they permit the up-and-down conversion of the 40 TV channels at the cost of a single channel transceiver and the modulating transform. Typical implementations rely on power-of two fast Fourier transforms (FFTs); however the smallest transform covering the full 40 channels of the TVWS spectrum is a 64 points wide, thus involving 24 unused channels. In order to attain a more numerically-efficient implemented design, we introduce the use of mixed-radix FFTs modulating transform. Testing various sizes and architectures, this approach provides up to 6.7% of energy saving compared to previous designs. Different from orthogonal frequency-division multiplexing (OFDM), FBMC systems are generally expected to be more robust to synchronisation errors, as oversampled FBMC systems can include a guard band, and even in a doubly-dispersive channel, inter-carrier interference (ICI) can be considered negligible. Even though sub-channels can be treated independently-i.e. without the use of cross-terms-they still require equalisation. We introduce a per-band equalisation, amongst different options, a robust and fast blind approach based on a concurrent constant modulus (CM)/decision directed (DD) fractionally-space equaliser (FSE) is selected. The selected approach is capable of equalising a frequency-selective channel. Furthermore the proposed architecture is advantageous in terms of power consumption and implementation cost. After focussing on the design of the radio for TVWS transmission, we address a multi-radio user assignment problem. Using various power consumption and harvesting models for the base station, we formulate two optimisation problems, the first focuses on the base station power consumption, while the second concentrates on load balancing. We employ a dynamic programming approach to optimise the user assignment. The use of such algorithms could
Date of Award22 Jun 2021
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SupervisorStephan Weiss (Supervisor) & Louise Helen Crockett (Supervisor)

Cite this

'