Reliability versus Cost in Next Generation Optical Access Networks

University dissertation from Stockholm : KTH Royal Institute of Technology

Abstract: The ever increasing demands of Internet users caused by the introduction of new high bandwidth applications and online services as well as the growing number of users and devices connected to the Internet, bring many challenges for the operators, especially in the last mile section of the network. Next generation access architectures are expected to offer high sustainable bandwidth per user. They also need to support a much larger service areas to decrease number of current central offices and hence potentially save the network expenditures in the future. Obviously, it requires high capacity and low loss transmission and optical fiber technology is the only future proof candidates for broadband access. Although this technology has already been widely deployed in the core networks, it is hard to use the same expensive devices made for core segment to solve the last mile bottlenecks, due to the low number of users sharing the network resources (and deployment cost). Therefore, the next generation optical access (NGOA) networks need to be designed with consideration of cost efficiency in the first place. Network reliability is also turning to be an important aspect for the NGOA networks as a consequence of long reach, high client count and new services requiring uninterrupted access. Consequently, new architectures not only need to be cost efficient but also they should fulfill the increasing reliability requirements. Although several NGOA alternatives have been proposed in the literatures, there is not yet an agreement on a single architecture. As described earlier, network expenditure and reliability performance are the two main factors to be considered. Therefore, this thesis concentrates on finding a suitable alternative for future broadband access by evaluating the reliability performance and total cost of ownership for several NGOA candidates. In particular, in this thesis we analyze the tradeoff between the cost needed to deploy backup resources and the reliability performance improvement obtained by the provided survivability mechanism. First, we identified the suitable NGOA candidates by comparing two main groups of optical access networks, namely passive optical networks (PONs) and active optical networks (AONs), in terms of cost, reliability performance and power consumption. The initial results have shown that wavelength division multiplexing PON (WDM PON) is the most promising alternative for the NGOA networks because of its high potential capacity, low cost and power consumption. So we continued our studies by investigating two WDM-based PON architectures regarding their cost and reliability performance. The study has also included a proposed fiber layout compatible with these two candidates aiming to minimize the required investment needed to offer protection. Our primary results confirmed that hybrid PON (HPON) is the best alternative for the NGOA networks. Therefore we further analyzed this candidate considering several variants of HPON. The most important components and sections of the HPON, which need to be protected to decrease the impact of each failure in the network have been identified. Based on these outcomes, two resilience architectures protecting the shared part of the HPON were proposed and their reliability performance parameters as well as cost of protection were evaluated. According to the results, using our proposed protection schemes a considerable improvement in reliability performance of the HPON variants can be provided at minor extra investment. We also introduced a cost efficient HPON architecture with different levels of protection for users with various reliability requirements, i.e. the protection of shared parts of the access network for all the connected users and end-to-end resilience scheme for some selected ones (e.g., business users). To gain an overall view on the cost efficiency of the proposed architecture, we evaluated the investment required for deploying these schemes considering several network upgrading paths towards a protected network. Moreover, a sensitivity analysis investigating the influence of network deployments time and the density of the users with higher availability requirements was presented. In summary, we have shown that HPON is able to fulfill the main NGOA requirements such as high bandwidth per-user, large coverage and client count. The work carried out in the thesis has proved that HPON can also offer high reliability performance while keeping the network expenditures at an acceptable level. Moreover, low power consumption and high flexibility in resource allocation of this architecture, makes it a winning candidate for the NGOA networks. Therefore, HPON is a promising architecture to be deployed as NGOA network in the near future considering the fact that components are soon to be available in the market.