Abstract
Networks and networking technologies are the key components of Big Data systems. Modern and future wireless sensor networks (WSN) act as one of the major sources of data for Big Data systems. Wireless networking technologies allow to offload the traffic generated by WSNs to the Internet access points for further delivery to the cloud storage systems. In this thesis we concentrate on the detailed analysis of the following two networking aspects of future Big Data systems: (i) efficient data collection algorithms in WSNs and (ii) wireless data delivery to the Internet access points.
The performance evaluation and optimization models developed in the thesis are based on the application of probability theory, theory of stochastic processes, Markov chain theory, stochastic and integral geometries and the queuing theory.
The introductory part discusses major components of Big Data systems, identify networking aspects as the subject of interest and formulates the tasks for the thesis. Further, different challenges of Big Data systems are presented in detail with several competitive architectures highlighted. After that, we proceed investigating data collection approaches in modern and future WSNs. We back up the possibility of using the proposed techniques by providing the associated performance evaluation results. We also pay attention to the process of collected data delivery to the Internet backbone access point, and demonstrate that the capacity of conventional cellular systems may not be sufficient for a set of WSN applications including both video monitoring at macro-scale and sensor data delivery from the nano/micro scales. Seeking for a wireless technology for data offloading from WSNs, we study millimeter and terahertz bands. We show there that the interference structure and signal propagation are fundamentally different due to the required use of highly directional antennas, human blocking and molecular absorption. Finally, to characterize the process of collected data transmission from a number of WSNs over the millimeter wave or terahertz backhauls we formulate and solve a queuing system with multiple auto correlated inputs and the service distribution corresponding to the transmission time over a wireless channel with hybrid automatic repeat request mechanism taken into account.
The performance evaluation and optimization models developed in the thesis are based on the application of probability theory, theory of stochastic processes, Markov chain theory, stochastic and integral geometries and the queuing theory.
The introductory part discusses major components of Big Data systems, identify networking aspects as the subject of interest and formulates the tasks for the thesis. Further, different challenges of Big Data systems are presented in detail with several competitive architectures highlighted. After that, we proceed investigating data collection approaches in modern and future WSNs. We back up the possibility of using the proposed techniques by providing the associated performance evaluation results. We also pay attention to the process of collected data delivery to the Internet backbone access point, and demonstrate that the capacity of conventional cellular systems may not be sufficient for a set of WSN applications including both video monitoring at macro-scale and sensor data delivery from the nano/micro scales. Seeking for a wireless technology for data offloading from WSNs, we study millimeter and terahertz bands. We show there that the interference structure and signal propagation are fundamentally different due to the required use of highly directional antennas, human blocking and molecular absorption. Finally, to characterize the process of collected data transmission from a number of WSNs over the millimeter wave or terahertz backhauls we formulate and solve a queuing system with multiple auto correlated inputs and the service distribution corresponding to the transmission time over a wireless channel with hybrid automatic repeat request mechanism taken into account.
| Original language | English |
|---|---|
| Publisher | Tampere University of Technology |
| Number of pages | 45 |
| ISBN (Electronic) | 978-952-15-3880-3 |
| ISBN (Print) | 978-952-15-3865-0 |
| Publication status | Published - 28 Nov 2016 |
| Publication type | G5 Doctoral dissertation (articles) |
Publication series
| Name | Tampere University of Technology. Publication |
|---|---|
| Volume | 1444 |
| ISSN (Print) | 1459-2045 |