File(s) under permanent embargo
Reason: Restricted by author. A copy can be supplied under Section 51(2) of the Australian Copyright Act 1968 by submitting a document delivery request through your library or by emailing email@example.com
Enhanced QoS centric backoff algorithm for wireless sensor network
thesisposted on 15.02.2017, 23:35 by Al-Mamun, Miraz
Wireless Sensor Network (WSN) is the state-of-the-art technology to perform distributed processing of data and form a synergy to accomplish intended functions. It has evolved from WLAN, WMAN and WMN to leverage the advantage of micro-processing, mobility and ubiquity to perform tasks such as accessibility to hard-reach area, deployment in extreme conditions, collection of data in geographically diversified fields and so on, whereas other types of wireless network have got constraints. In a distributed node structure like WSN, one of the pivotal concerns of exchanging data is the ability to streamline data access in between the nodes when all the nodes demand for share of channel and avoid bottleneck and high-latency as much as possible. If the radio channel itself is spectrum constrained e.g. single-channel radio, the nodes must operate in a fashion to minimize the collision of accessing shared channel. This issue severely affects the network throughput when packets of different classes (e.g. audio, video, general data, network status and so on) are handled. In this scenario, random channel access mechanism enables the allocation of radio channel to the nodes and streamline the packet transfer. Existing random channel access mechanism makes the sender node to wait for a random period out of predefined waiting time window which is called backoff period, before acquiring the channel and avoid collision. But, this mechanism implements uniform distribution of selecting this random waiting time out of the backoff period. This causes the overlap of the selected time and results in collision. In our thesis work we propose, non-uniform probability distribution of selecting random waiting period. This method segments the whole backoff period in small time slots and augments the probability of sending data in a time slot following an exponential pattern as each of the previous time slots go silent (i.e. none of the senders contending does access the channel). This method gives fine grained control of the channel access and allows the mechanism to reflect Quality of Service (QoS) as the probability of accessing the channel can be optimized by custom parameters. Our proposed method has been tested by simulation environment in NS 2, one of the state-of-the-art network simulation environments. The tests results speak for themselves to show better latency, throughput and less collision.