Dynamic Bandwidth Allocation Algorithm using Sequential Channel Selection for Hybrid WDM/TDM EPON

Abstract

The optical communication infrastructure has introduced a revolutionary growth in the backbone, resulting the emergence of high bandwidth requiring applications. The current use of cable modem and digital subscriber loop in access premises are gradually becoming unable to cope with the increased bandwidth demanding application, whilst the optical infrastructure can support these in backbone, wide area networks and metropolitan area networks. Passive Optical Network (PON) has emerged a promising solution to wipe out the bandwidth bottle neck offered by copper/coax communication infrastructure. The key PON variants are, Time Division Multiplexing (TDM) PON in which the bandwidth is shared among multiple users and Wavelength Division Multiplexing (WDM) PON where each user enjoys complete bandwidth pipe offered by a wavelength channel. Upgrading the current generation TDM PON to the future WDM PON would be a challenge when the end users demand outgrows current TDM network bandwidth. This thesis presents a compromised migration path from current TDM PON to future WDM PON by exploiting an intermediate solution Hybrid WDM/TDM PON. The proposed network has been anticipated as NUST Hybrid (NH) PON (i.e., an access solution for NUST H-12 campus). The dual ring NH-PON meets the significant challenges of network resilience and high scalability. Bandwidth allocation having dynamic aspects for Hybrid WDM/TDM PON is also a part of this research. Thesis proposes a dynamic bandwidth allocation (DBA) algorithm with sequential channel selection (SCS) keeping the scalability, fairness, simplicity (i.e., not to over load OLT s performance) and robustness issues upfront for Hybrid WDM/TDM PON. The main focus is to amalgamate the pros of bandwidth allocation schemes of key xiii PON variants (i.e., TDM PON and WDM PON). DBA for TDM PON has emerged quite maturely yet finds major draw back as the network size increases. DBA in WDM PON (in contrast to TDM PON) does not find the bottle neck of increased scalability, yet the over all network performance is degraded as some wavelengths are over loaded while others under loaded. The main idea of SCS is to serve more number of TDM PONs (i.e., increased scalability) by allocating bandwidth dynamically with lesser tunable transceivers (i.e., reducing the overall cost of transceivers in the network) at the OLT, where each attached PON is being served over distinct wavelength. Wavelength channels for each PON are allocated statically and the OLT tunes its transceivers to each channel in round robin fashion providing bandwidth to each PON dynamically. The SCS has been evaluated for performance parameters as average packet delay and mean queuing depth. Results show that SCS outperforms static bandwidth allocation for both performance evaluation parameters.