PhD Thesis

  • Title:
    Cognitive Radio In HF Communications: Selective Transmission and Broadband Acquisition


Date: 23 May 2016

Author: Laura Beatriz Melián Gutiérrez

Directors:

  • Dr. Iván Alejandro Pérez Álvarez
  • Dr. Santiago Zazo Bello


Presentation of Thesis

     portada presentacion tesis


Thesis

        logo pdf

Abstract:

The HF band (3-30 MHz) promotes the establishment of trans-horizon radio links by using the ionosphere as a passive reflector. Although HF radio links have this transhorizon behaviour, most HF stations are allocated by national regulators. This allocation results in multiple collisions between HF users even if they are legacy users in their respective countries. Furthermore, HF communications make use of the Automatic Link Establishment (ALE) protocol, which has been referred to as an example of a primitive form of cognitive radio. It is based on a listen-before-transmit strategy to avoid interfering with on-going communications. However, the ALE protocol has several limitations: it does notmanage the spectrumin a wide-band sense and it does not monitor the evolution of the users’ activity in the channels. Due to these limitations, more dynamic techniques than the ALE protocol must be implemented in HF stations to reduce the amount of collisions between users. New capabilities such as adaptability and cognition have to be introduced in HF stations to reduce the inefficient use of this band in terms of successful access to the spectrum resources.

In this Thesis, the application of cognitive radio principles is proposed to reduce the amount of collisions between HF users and to reduce the inefficient use of this band. The cycle of tasks that a cognitive radio should face can be divided into three main tasks: Observe, Learn, and Decide & Act. They represent the cycle from the spectrum acquisition to the selection of the best channel to transmit, according to the observed and learned activity of other users.

A database of real measurements of the activity in the HF band is created in order to evaluate and validate the proposed cognitive techniques in this Thesis. This database contains wideband measurements of the power spectrum of the HF band. By using an energy detector to perform the spectrum sensing task, the acquired power spectrum is converted into users’ activity information. The energy detector is the most appropriate technique in such a heterogeneous environment as the HF band.

One of the challenges that the acquisition of wideband measurements faces is the effect of narrowband interference (NBI) in wideband receivers. NBI causes a reduction in the effective number of bits used for the digitalisation of the signals of interest, and the quantization noise can exceed the thermal noise and the desired signal itself. Since NBI has to be mitigated in the analog domain, a compressive sensing based NBI detector is proposed to identify NBI before the digital front-end in wideband HF receivers.

Two learning strategies are defined and validated in this Thesis. An activity model based on Hidden Markov Models (HMM) is designed and validated for long-term predictions of the activity in a particular HF channel. Another learning strategy is validated for short-term predictions in HF channels, the Upper Confidence Bound (UCB) algorithm. Besides its use for short-term predictions it also allows for decision-making to select the best channel to transmit in terms of availability.

Finally, a hybrid system combining previous learning strategies is defined in this Thesis. This hybrid UCB-HMM system can be seen as a metacognitive engine, which is able to adapt its data transmissions, i.e. it is able to select the most appropriate cognitive engine to transmit, according to the observed changes in the environment. Besides its adaptability to the changes in the environment, it is also shown that the amount of signalling information exchanged between transmitter and receiver is significantly decreased.