Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/84463
Title: The modeling and optimization of 802.11p VANETs unicast performance
Authors: Xie, Yu
Degree: M.Phil.
Issue Date: 2017
Abstract: 802.11p vehicular ad-hoc networks (VANETs) are drawing growing research attentions, as it will play an important role in future Intelligent Transportation Systems (ITS) for ubiquitous communications and connectivity of vehicles. Various messages can be transmitted in a VANET to improve road safety and furnish multiple types of application services. Therefore, the evaluation of VANET performance and its optimization should be indispensably considered. Previous conventional considerations of pertinent studies on VANET modeling did not take a realistic vehicular traffic distribution into account. They merely incorporated a general homogeneous road scenario. Furthermore, most of previous works primarily focused on the broadcasting performance in VANETs, since the safety beacon packets, which were crucial for reducing traffic accidents, were transmitted in periodic broadcast. However, with respect to certain service data (e.g., sensor data), unicast with the re-transmission mechanism is more appropriate, as it aims to ensure successful reception of data. Considering the prospective integration of VANET with the mobile Internet, unicast should also be brought to the forefront. On the other hand, most of the VANET optimization schemes amid previous researches required continuous monitoring of the network and measuring the number of neighboring nodes (e.g., through the feedback-loop principle or local neighbor discovery) to configure the transmission power or adjusting the transmission rate accordingly. Such monitoring and measurement led to large transmission overheads and measurement delay. In view of these inadequacies, a set of 802.11p unicast modeling and optimization methodologies without continuous measuring the number of neighbors are proposed under a practical stochastic traffic modeling framework for estimating and optimizing the vehicle-to-vehicle (V2V) network performance in this thesis.
The modeling is composed of four portions: (i) a stochastic traffic model that describes the realistic traffic road with traffic signaling lights and outputs a vehicular density profile based on the empirical velocity profile; (ii) the contention model based on a two-dimensional Markov chain depicting the 802.11p unicast channel access contention process of each node at different locations on the road with the density profile from the stochastic traffic model; (iii) an interference model which characterizes the interference triggered by concurrent transmissions and hidden nodes to each node in the network with the foregoing density profile; and (iv) the performance model that analyzes the delay and throughput performances for each node at dissimilar locations based on the resultant parameters output from the two foresaid models. In sum, given a velocity profile as the input, the analytical delay and throughput performance can be attained through our modeling. The feasibility of these modeling methodologies has been rigorously verified by extensive simulation. In both the analytical and simulated results for delay and throughput, we found that the signal-controlled stochastic traffic distribution of vehicles inflicts conspicuous impact on the unicast performance, which provides insights into the studies of the interaction between road traffic and communication network performance. The modeling methodologies proposed in this thesis can be utilized to predict network performance, and traffic and network planning can be carried out respectively to further optimize the data delivery in VANETs. Based on the analytical modeling of 802.11p VANETs performance, state-of-the-art optimization methods are put forward for each node in the network to reduce packet collisions and enhance the overall network performance. For instance, the optimal transmission range and the optimal Medium Access Control (MAC) contention window size at different locations can be derived prior to vehicles' entering the road segment from the established unicast modeling methodology. With the optimal values reaped pre-emptively, vehicles can adjust their relevant performance parameters once they arrive at corresponding locations on the road. The cross-layer proposal in this thesis involves both the physical (PHY) layer controlling the transmission power and the MAC layer controlling the channel access rate, and the optimization schemes are well-evaluated through extensive simulation. Our results indicate that delay (throughput) is improved by about 53% (120%) on average for homogeneous traffic with different density levels and 45% (104%) on average for heterogeneous traffic at all locations through the cross-layer optimization.
Subjects: Vehicular ad hoc networks (Computer networks)
Hong Kong Polytechnic University -- Dissertations
Pages: 94 pages : color illustrations
Appears in Collections:Thesis

Show full item record

Page views

55
Last Week
1
Last month
Citations as of Mar 24, 2024

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.