Cooperative spectrum sensing based on machine learning in cognitive radio vehicular network / Mohammad Asif Hossain

Mohammad Asif , Hossain (2022) Cooperative spectrum sensing based on machine learning in cognitive radio vehicular network / Mohammad Asif Hossain. PhD thesis, Universiti Malaya.

[img] PDF (The Candidate's Agreement)
Restricted to Repository staff only

Download (229Kb)
    [img] PDF (Thesis PhD)
    Restricted to Repository staff only until 31 December 2023.

    Download (4Mb)


      A vehicular ad hoc network (VANET) is a major participant in reducing traffic collisions and road congestion. In VANET, an enormous amount of data is needed to be exchanged. However, the dedicated access technique for VANET, the DSRC (dedicated short-range communication), is insufficient to accommodate these enormous data quantities. Consequently, VANET faces a lack of spectrum bands. Cognitive radio (CR) can be used to alleviate such a crisis. CR is an intelligent and programmable radio device that accesses various frequency bands. To enter these bands, CR must conduct spectrum sensing. An SU (secondary user) or a CR user can decide whether the spectrum is occupied by some PUs (primary or licensed users) or not. This mechanism is known as spectrum sensing. It is the technique to identify any vacant spectrum (no licensed user is currently using). When it discovers that the PU does not use the licensed spectrum, SU will use it (vacant licensed spectrum) under some restrictions when the DSRC is filled. Several spectrum sensing techniques are found in the literature. Few examples are energy detection, cyclostationary detection, matched filter detection, etc. Due to the speedy nature of the vehicles, these spectrum sensing techniques are confronted with several problems such as rapid change of radio environment, heterogeneous service quality (QoS) requirements, hidden node problem, problems in multipath fading and shadowing, diffraction, various PU activity models, etc. Spectrum sensing must be accurate, faster, dynamic, and flexible due to vehicles' high velocity and constant environmental shift. The optimum spectrum sensing should consider vehicle speed, fading degree, density, radio environment, etc. To solve the above-mentioned issues and problems, this thesis presents a novel segment-based CR-VANET (Seg-CR-VANET) cooperative spectrum sensing (CSS) framework. In this thesis, roads are divided into equal segments, which are then subdivided depending on the probability value. CSS needs to perform two sensing for its operation; local sensing and global sensing. In this thesis, local sensing results are produced by SUs or vehicles by selecting the best spectrum sensing technique. The selection would be made based on the hybrid machine learning (ML) algorithm. A fuzzy-based Naive Bayes algorithm has been used in this case. For the local sensing, values of signal-to-noise ratio (SNR) and noise uncertainty are used. Instead of using a static threshold value, dynamic threshold values are used for sensing techniques. The tri-agent reinforcement learning (TA-RL) algorithm has been used by the segment spectrum agent (SSA) in this suggested CSS to make the global decision. TA-RL algorithm learns three environments (network, signal, and vehicle behavior) to assess the behaviors of PUs. The whole solution has been implemented in OMNeT++ with the required modules and the SUMO traffic generator tool. The simulation findings reveal that the Seg-CR-VANET outperforms existing works in terms of several performance metrics used in spectrum sensing such as probability of detection, false alarm rate, accuracy, throughput, delay, packet delivery ratio, etc.

      Item Type: Thesis (PhD)
      Additional Information: Thesis (PhD) – Faculty of Computer Science & Information Technology, Universiti Malaya, 2022.
      Uncontrolled Keywords: Spectrum sensing; Cognitive radio; VANET; Tri-agent reinforcement learning; Machine learning; Dynamic threshold values; Road segmentation
      Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
      Q Science > QA Mathematics > QA76 Computer software
      Divisions: Faculty of Computer Science & Information Technology
      Depositing User: Mr Mohd Safri Tahir
      Date Deposited: 06 Apr 2023 07:02
      Last Modified: 06 Apr 2023 07:02

      Actions (For repository staff only : Login required)

      View Item