Electrical and Electronic Engineering - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemNext-generation energy-efficient broadband access networks( 2016)The popularity of bandwidth-intensive applications and the deployment of virtual private networks (VPNs) have resulted in a rapid growth of the Internet during recent years. The demand for high bandwidths and the necessity to serve customers over an extended reach cannot be met with conventional wired broadband access technologies, such as digital subscriber line (DSL), due to bandwidth limitations of the copper medium. To overcome this bottleneck and deliver high-speed content to customers over various distances, deploying optical fibre in the access network has been proposed as the most appropriate strategy. After considering potential optical fibre configurations for the amount of fibre and the number of transceivers required, the passive optical network (PON) has been selected as the most efficient architecture to provide fibre access to its customers. Today, PONs are not only used as access networks for fixed wire-line customers, but also as backhaul networks for mobile users. PONs are more energy efficient than their copper counterparts as a result of low transmission losses, high bandwidths, and passive components present. Due to their ability to deliver high bandwidths in a cost-effective manner, a large number of PONs have been deployed globally. Wider deployment of PONs increases the number of customers and user equipment in communication networks, thereby increasing the overall energy consumption. The implications of increased energy consumption, such as increased greenhouse gas emissions and operational expenditure, have led to the emergence of energy conserving studies for PON. The main objective of this thesis is to propose energy-efficient algorithm-based solutions for the PON. For this purpose, we have selected the time division multiplexed (TDM)-PON and the time and wavelength division multiplexed (TWDM)-PON, as they are considered to be the most popular PON choices for current and future high-speed communication networks, respectively. In both TDM-PON and TWDM-PON, each optical network unit (ONU) is allocated a time slot to access the upstream and downstream bandwidths. Beyond this time slot, an ONU remains idle, i.e., does not exchange or process packets. Despite being idle, ONUs continue to operate at their active power level. As a result, the high energy consumption of the PON is attributed to the ONUs, which continuously operate at their active power level and are present in large numbers in current communication networks. As such, minimising ONU energy consumption has been identified as the best approach to reduce overall energy consumption of the PON. For this purpose, transitioning idle ONUs into low-power modes, such as sleep and doze, have been proposed for both TDM-PON and TWDM-PON. In sleep mode, both the ONU transmitter (Tx) and receiver (Rx) are powered down, while in doze mode, only ONU Tx is powered down. Moreover, multiple wavelengths and tunable ONU transceivers (TRXs) present in TWDM-PONs facilitate further energy savings at the OLT through wavelength reallocation. Wavelength reallocation involves switching off idle wavelengths and redistributing the ONUs amongst remaining active wavelengths. In this thesis, we exploit the sleep/doze capabilities and wavelength tunability of the ONUs in designing energy-efficient dynamic bandwidth allocation (DBA) algorithms and dynamic wavelength and bandwidth allocation (DWBA) algorithms to minimise the energy consumption of the TDMPON and TWDM-PON, respectively. Proposing sleep/doze-enabled DBA algorithms for the TDM-PON involves addressing key operational and performance challenges that arise in the network due to sleep and doze mode operations. First, sleep/doze mode operations should be incorporated into the underlying DBA algorithm of the network. This thesis investigates how generalMACcontrol messages can be used to exchange sleep/doze control messages between the OLT and ONUs. Another design parameter, sleep or doze duration, is carefully examined in this thesis as it affects both energy savings and quality of service (QoS). Longer sleep/doze duration improves energy savings at the expense of increased average delay. As a result, energy consumption and average delay cannot be minimised simultaneously. To this end, this thesis explores sleep/doze control functions that adapt to varying traffic loads of the network to reduce energy consumption and also average delay. We also investigate supporting estimation techniques used in these sleep/doze control functions. Finally, for the TDM-PON, we investigate possible methods of minimising the QoS degradation due to sleep/doze mode operations. In TWDM-PON, both sleep/doze mode operations and wavelength reallocation can be implemented to minimise energy consumption of the ONUs and OLT, respectively. The DWBA algorithms of this nature address two key challenges, namely, determining the number of active wavelengths and sustaining QoS specifications of the network. In current customer-centric telecommunication networks, network operators are bound to deliver a satisfactory QoS to customers through service level agreements (SLAs). As failure to comply with SLAs results in heavy penalties for network operators, meeting QoS requirements of the network is important to generate positive revenue. However, they also face the challenge of reducing the energy consumption of the network to meet environmental standards as well as to reduce the operational expenditure. To this end, we formulate models that determine the number of active wavelengths to improve energy savings and also to satisfy application-specific delay constraints. The performance of different types of DWBA algorithms is also evaluated under the proposed model. Finally, the performance of our proposed solutions is analysed for various types of network traffic. Overall, the technical contributions presented in this thesis provide insight into how power saving techniques can be incorporated into the underlying DBA and DWBA algorithms to improve energy savings and sustain delay requirements of the PON.We also discuss how these solutions can be extended to different scenarios and applications in the future.