Network and service monitoring in heterogeneous home networks

A. Delphinanto

Research output: ThesisPhd Thesis 1 (Research TU/e / Graduation TU/e)

797 Downloads (Pure)

Abstract

Home networks are becoming dynamic and technologically heterogeneous. They consist of an increasing number of devices which offer several functionalities and can be used for many different services. In the home, these devices are interconnected using a mixture of networking technologies (for example, Ethernet, Wifi, coaxial cable, or power-line). However, interconnecting these devices is often not easy. The increasing heterogeneity has led to significant device- and service-management complexity. In addition, home networks provide a critical "last meters" access to the public telecom and Internet infrastructure and have a dramatic impact on to the end-to-end reliability and performance of services from these networks. This challenges service providers not only to maintain a satisfactory quality of service level in such heterogeneous home networks, but also to remotely monitor and troubleshoot them. The present thesis work contributes research and several solutions in the field of network and service monitoring in home networks, mainly in three areas: (1) providing automatic device- and service-discovery and configuration, (2) remote management, and (3) providing quality of service (QoS). With regard to the first area, current service discovery technology is designed to relieve the increasing human role in network and service administration. However, the relevant Service Discovery Protocols (SDPs) are lacking crucial features namely: (1) they are not platform- and network-independent, and (2) they do not provide sufficient mechanisms for (device) resource reservation. Consequently, devices implementing different SDPs cannot communicate with each other and share their functionalities and resources in a managed way, especially when they use different network technologies. As a solution to the first problem, we propose a new proxy server architecture that enables IP-based devices and services to be discovered on non-IP based network and vice versa. We implemented the proxy architecture using UPnP respectively Bluetooth SDP as IP- and non-IP-based SDPs. The proxy allows Bluetooth devices and UPnP control points to discover, access, and utilize services located on the other network. Validation experiments with the proxy prototype showed that seamless inter-working can be achieved keeping all proxy functionalities on a single device, thus not requiring modification of currently existing UPnP and Bluetooth end devices. Although the proxy itself taxes the end-to-end performance of the service, it is shown to be still acceptable for an end user. For mitigating resource conflicts in SDPs, we propose a generic resource reservation scheme with properties derived from common SDP operation. Performance studies with a prototype showed that this reservation scheme significantly improves the scalability and sustainability of service access in SDPs, at a minor computational cost. With regard to the second area, it is known that the end-to-end quality of Internet services depends crucially on the performance of the home network. Consequently, service providers require the ability to monitor and configure devices in the home network, behind the home gateway (HG). However, they can only put limited requirements to these off-the-shelf devices, as the consumer electronics market is largely outside their span of control. Therefore they have to make intelligent use of the given device control and management protocols. In this work, we propose an architecture for remote discovery and management of devices in a highly heterogeneous home network. A proof-of-concept is developed for the remote management of UPnP devices in the home with a TR-069/UPnP proxy on the HG. Although this architecture is protocol specific, it can be easily adapted to other web-services based protocols. Service providers are also asking for diagnostic tools with which they can remotely troubleshoot the home networks. One of these tools should be able to gather information about the topology of the home network. Although topology discovery protocols already exist, nothing is known yet about their performance. In this work we propose a set of key performance indicators for home network topology discovery architectures, and how they should be measured. We applied them to the Link-Layer Topology Discovery (LLTD) protocol and the Link-Layer Discovery Protocol (LLDP). Our performance measurement results show that these protocols do not fulfill all the requirements as formulated by the service providers. With regard to the third area, current QoS solutions are mostly based on traffic classification. Because they need to be supported by all devices in the network, they are relatively expensive for home networks. Furthermore, they are not interoperable between different networking technologies. Alternative QoS provision techniques have been proposed in the literature. These techniques require end-user services to pragmatically adapt their properties to the actual condition of the network. For this, the condition of the home network in terms of its available bandwidth, delay, jitter, etc., needs to be known in real time. Appropriate tools for determining the available home network resources do not yet exist. In this work we propose a new method to probe the path capacity and available bandwidth between a server and a client in a home network. The main features of this method are: (a) it does not require adaptation of existing end devices, (b) it does not require pre-knowledge of the link-layer network topology, and (c) it is accurate enough to make reliable QoS predictions for the most relevant home applications. To use these predictions for effective service- or content-adaptation or admission control, one should also know how the state of the home network is expected to change immediately after the current state has been probed. However, not much is known about the stochastic properties of traffic in home networks. Based on a relatively small set of traffic observations in several home networks in the Netherlands, we were able to build a preliminary model for home network traffic dynamics.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Electrical Engineering
Supervisors/Advisors
  • Koonen, A.M.J. (Ton), Promotor
  • Liotta, Antonio, Promotor
  • Hartog, den, F.T.H., Copromotor, External person
Award date15 Feb 2012
Place of PublicationEindhoven
Publisher
Print ISBNs978-90-386-3094-6
DOIs
Publication statusPublished - 2012

Fingerprint

Dive into the research topics of 'Network and service monitoring in heterogeneous home networks'. Together they form a unique fingerprint.

Cite this