Jawwad Ahmed

A comparison of dynamic bandwidth allocation for EPON, GPON, and next-generation TDM PON

Dynamic bandwidth allocation in passive optical networks presents a key issue for providing efficient and fair utilization of the PON upstream bandwidth while supporting the QoS requirements of different traffic classes. In this article we compare the typical characteristics of DBA, such as bandwidth utilization, delay, and jitter at different traffic loads, within the two major standards for PONs, Ethernet PON and gigabit PON. A particular PON standard sets the framework for the operation of DBA and the limitations it faces. We illustrate these differences between EPON and GPON by means of simulations for the two standards. Moreover, we consider the evolution of both standards to their next-generation counterparts with the bit rate of 10 Gb/s and the implications to the DBA. A new simple GPON DBA algorithm is used to illustrate GPON performance. It is shown that the length of the polling cycle plays a crucial but different role for the operation of the DBA within the two standards. Moreover, only minor differences regarding DBA for current and next-generation PONs were found.

Dynamic bandwidth allocation for long-reach PON: overcoming performance degradation

A passive optical network, with its inherent point to multi-point structure, allows for centralized placement of active equipment and possible extension of its boundary towards core networks. This property of the PON can be exploited for node consolidation where multiple central offices are replaced by a single one covering a larger service area. Such node consolidation is being particularly driven by the need for network operational cost saving, and is offering significant challenges to PONs. The degree of node consolidation that can be achieved is limited by the reach of conventional PON systems. In order to achieve a larger degree of node consolidation, an extension of the PON reach, beyond the conventional 20 km, is required. This article addresses the challenges of the dynamic bandwidth allocation, where increased reach results in a degradation of DBA performance and quality of service support. This degradation is a consequence of the increased propagation delay of the DBA messages exchanged between different PON elements. A potential solution to the performance degradation is the introduction of a multi-threaded DBA. In this article, we examine for both Gigabit PON and Ethernet PON, the extent to which DBA performance degradation can be reduced by exploiting multi-threading. It is found that for both standards, multi-threading, if done properly, can be used to mitigate the performance degradation due to the increased reach. To make bandwidth allocation efficient, new schemes for coordinating the multiple threads are required in long reach PON.

Efficient inter-thread scheduling scheme for long-reach passive optical networks

These days there is a clear trend toward extending the reach of passive optical networks to cover large geographical areas, which enables reduction of the number of central offices and hence has the potential of cost saving in network operation. On the other hand, this reach extension necessitates the design of efficient dynamic bandwidth allocation schemes in order to tackle performance degradation caused by the increased propagation delay in long reach PONs. Among many existing approaches, the multithread-based DBA scheme where several bandwidth allocation processes are performed in parallel is considered one of the most effective options to improve network performance in LRPONs. We have found that without proper intercommunication between the overlapped threads, multi-thread DBA may lose efficiency and even perform worse than the conventional singlethread algorithm. With this in mind, this article reviews different inter-thread scheduling schemes for LR-PONs, and proposes a novel approach of integrating the key ideas of the existing ones. Extensive simulation results confirm that our proposed scheme can significantly improve DBA performance for LR-PONs under a variety of scenarios with consideration of different values of network load and reach.

Predicting real-time service-level metrics from device statistics

While real-time service assurance is critical for emerging telecom cloud services, understanding and predicting performance metrics for such services is hard. In this paper, we pursue an approach based upon statistical learning whereby the behavior of the target system is learned from observations. We use methods that learn from device statistics and predict metrics for services running on these devices. Specifically, we collect statistics from a Linux kernel of a server machine and predict client-side metrics for a video-streaming service (VLC). The fact that we collect thousands of kernel variables, while omitting service instrumentation, makes our approach service-independent and unique. While our current lab configuration is simple, our results, gained through extensive experimentation, prove the feasibility of accurately predicting client-side metrics, such as video frame rates and RTP packet rates, often within 10–15% error (NMAE), also under high computational load and across traces from different scenarios.

A Dynamic Bulk Provisioning Framework for Concurrent Optimization in PCE-Based WDM Networks

A centralized network control and management plane, such as the one based on a path computation element (PCE), is highly beneficial in terms of resource optimization in wavelength division multiplexing optical networks. Benefits of centralized provisioning are even more evident when connection requests are provisioned in batches, i.e., they allow a better use of network resources via concurrent optimization. In this study, a dynamic bulk provisioning framework is presented with the objective of optimizing the use of network resources that also presents, as an additional benefit, the ability to yield a reduction of the control plane overhead. The rationale behind the proposed framework is based on a mechanism in which the PCE client is allowed to bundle and simultaneously send multiple labeled switch path (LSP) requests to the PCE where, in turn, several bundles can be concurrently processed together as a single bulk. From the network deployment perspective, a PCE-based network architecture is proposed to practically realize this approach. For dynamic bulk provisioning of optical LSP requests, a time-efficient integer linear programming (ILP) model (LSP_ BP_ ILP) is presented to minimize the request blocking, the network resource consumption, and the network congestion. In addition, a heuristic based on a greedy randomized adaptive search procedure (GRASP), namely LSP_BP_GRASP, is also proposed as a scalable alternative. The presented results demonstrate significant advantages of the proposed PCE bulk provisioning framework based on concurrent optimization in terms of reduced blocking probability and control overhead when compared with conventional dynamic connection provisioning approaches processing a single connection request at a time.

Predicting service metrics for cluster-based services using real-time analytics

Predicting the performance of cloud services is intrinsically hard. In this work, we pursue an approach based upon statistical learning, whereby the behaviour of a system is learned from observations. Specifically, our testbed implementation collects device statistics from a server cluster and uses a regression method that accurately predicts, in real-time, client-side service metrics for a video streaming service running on the cluster. The method is service-agnostic in the sense that it takes as input operating-systems statistics instead of service-level metrics. We show that feature set reduction significantly improves prediction accuracy in our case, while simultaneously reducing model computation time. We also discuss design and implementation of a real-time analytics engine, which processes streams of device statistics and service metrics from testbed sensors and produces model predictions through online learning.

Hybrid survivability schemes achieving high connection availability with a reduced amount of backup resources [invited]

Maximizing connection availability in wavelength division multiplexing (WDM) networks is critical because even small disruptions can cause huge data losses. However, there is a trade-off between the level of network survivability and the cost related to the backup resources to be provided. One-hundred percent survivability can be achieved by dedicated path protection (DPP) with multiple prereserved protection paths for each provisioned connection, i.e., DPP (1:N). Unfortunately, the blocking probability performance of DPP (1:N) is negatively affected by the large number of prereserved backup wavelengths standing by unutilized. On the other hand, path restoration (PR)-based solutions ensure good blocking performance at the expense of lower connection availability. The work in this paper aims at finding hybrid network survivability strategies that combine the benefits of both techniques (i.e., high availability with low blocking rate). More specifically, the paper focuses on a double link failure scenario and proposes two strategies. The first one couples DPP (1:1) with path restoration (referred to as DPP + PR) to minimize the number of dropped connections. The second scheme adds the concept of backup reprovisioning (BR), referred to as DPP + BR + PR, in order to further increase the connection availability achieved by DPP + PR. Integer linear programming models for the implementation of the proposed schemes are formulated. Extensive performance evaluation conducted in a path-computation-element-based WDM network scenario shows that DPP + BR + PR and DPP + PR can significantly lower the blocking probability value compared to DPP (1:2) without compromising too much in terms of connection availability.

Concurrent processing of multiple LSP request bundles on a PCE in a WDM network

Concurrent RWA algorithm for differentiated services to process multiple LSP bundles at PCE is proposed. Significant blocking probability reduction has been observed at the expense of slightly increased LSP setup-time compared to a sequential approach.

Lsp request bundling in a PCE-based WDM network

LSP requests may be bundled to improve the network optimization process at the expense of an increased connection setup delay. A detailed study is conducted to evaluate the pros and cons of the bundling approach.

Enhanced IPACT: solving the over-granting problem in long-reach EPON

In this paper we address an issue referred to as “over-granting problem”, which is inherent in the existing dynamic bandwidth allocation algorithms for Ethernet-based Passive Optical Networks (EPON), in particular when deployed for multi-threaded scheme in long-reach scenario. In order to solve this problem we design a scheme for the algorithm of Interleaved Polling with Adapted Cycle Time (IPACT) with the limited service discipline. We evaluate the proposed scheme through simulations for single-thread and double-thread cases and demonstrate that, the network performance can be significantly improved by our solution in terms of average delay, jitter, and throughput.