الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 86 |  |  |
| | | from | 86 | | |
Abstract
Wireless Sensor Network (WSN) is typically made up of small devices called sensors that have low battery power, limited processing capacity, limited storage capacity and are dispersed in a geographical area for sensing the environment and collecting data. Sensors, whether deployed in WSNs or Internet of Things (IoT), are embedded within objects to sense and collect physical phenomenon from the world. Each set of sensors forms a cluster with a cluster head (CH) which is responsible for collecting data from the cluster members (CMs) and sending it to a sink node (also called base station). The aggregated data are locally filtered and stored in a storage module called buffer. Then, data is conveyed over a network for storing, analyzing and processing in powerful fog and cloud servers. Since sensor nodes can be act as source nodes, routers and processing units, the energy represents a bottleneck in the WSN. Due to the massive number of sensor nodes, the data at CMs, CHs and sink nodes becomes very vast which resulted in buffer overflow and more packets get blocked or dropped. As the buffers get full, the number of retransmissions increases. As a result, sensor nodes lose significant amounts of energy during traffic transmissions. Buffer management and data reduction in wireless sensor networks (WSNs) are critical wherever buffer overflow and number of transmissions cause power waste and data loss. To improve energy consumption, this thesis presents an Energy-Efficient Buffer Management based on data integrity and multivariate data reduction (EEBM-IMR) scheme which operates at both the cluster heads and the sink nodes sides. To save the buffer space at iii the cluster heads, EEBM-IMR classifies the data measured by sensors, based on its integrity, into malicious or verified packets. To reduce data transmissions (i.e. transmission ratio), a multivariate data reduction scheme is introduced based on a binary tree data structure. The efficiency of EEBM-IMR is evaluated in terms of transmission ratio, dropping probability and throughput using a real world dataset of fifty sensors. The experimental results show that, as the number of sensors increases, EEBM-IMR saves energy and outperforms some existing models in previous studies.