Home networking, Residential gateways: the carrot and the stick
Suppliment July August 2008
The home gateway opens up a new front for operators, but most think it is too early and the risk too great for broadcast services, says Philip Hunter
Operators have been treading warily towards the home gateway, aware that the risk of brand damage through poor quality of service (QoS) is as great as the opportunity to tap new sources of revenue. But the day of reckoning can be delayed a little longer, for the technologies are falling into place for delivering high definition video within the home. Yet standards remain unclear and fragmented, while the tools for device and traffic management are still in their infancy. Some operators believe it is too early yet to take the plunge into full broadcast delivery within the home, through fear of losing hard won reputations for high quality, according to John Cassidy, a senior manager in the communications and high technology practice at global IT services and outsourcing company Accenture.
“As communication and entertainment providers start to manage more of the functions of the home they start to place their brand in the line of fire for things they do not control,” says Cassidy. The risk is that operators will assume responsibility for the home network while lacking sufficient control over it, with the onus on the gateway to provide visibility for management of traffic and devices on the other side. “While there is a great opportunity for the content and broadcast industry to step up and deliver high quality services within the home, opening up great new streams of revenue, history suggests that there will be severe teething problems,” suggests Cassidy. “The industry track record of doing this when they control all of the constituent parts is poor. Now, when more is outside their control there is a real risk of all the good work done to improve things over the last five years being un-done.”
Naturally vendors themselves are more optimistic, pointing out that the home gateway is not entirely uncharted territory, for similar ground has been covered already in the enterprise. The residential gateway can be regarded as a re-invention of the Integrated Access Device (IAD), which emerged during the 1990s as a common entry point to voice and data services for both small and medium enterprises (SMEs) and branch offices of large organisations, according to
Sanjeev Challa, chief technologist in the Gateway Products Group at Ikanos Communications, a maker of chips for broadband systems. But this analogy can only be used as a starting point, for, as Challa agrees, the home gateway will be a far more complex device than the IAD, having to support a greater range of physical media on the home network side especially, but above all running multimedia applications, while being responsible for both QoS and remote management.
Standards hinder the evolutionary path
It all started with the DSL modem, which evolved into a router and gained wireless capabilities with the advent of multiple PCs and WiFi. Up to this point the home gateway had indeed taken a similar path to the IAD, except for the wireless, but then started to incorporate techniques for QoS and traffic management to support voice over IP (VoIP). Now with the addition of video, QoS becomes more difficult to guarantee because of the much greater bandwidth needed, although the various physical home network types have all stepped up to the mark by supporting increased bit rates in their latest versions, typically around 200 Mbps in theory, even if the reality is rather different. But added to this is the spectre of fixed mobile convergence with IP Multimedia System (IMS), leading to integration between cellular and packet based services, as well as an emerging trend towards consolidated storage within the home for data, images, and video. This means the gateway will manage Network Attached Storage (NAS) for the home network and provide virtual PVR functionality for every TV.
Again unlike the IAD, the home gateway field has been handicapped by a plethora of standards operating at various levels, and uncertainty over which will prevail. First of all we have the physical home network itself, with four serious contenders. These are broadband powerline over power cables (BPL), telephone wiring, category 5/6 cabling for Ethernet, coaxial cable similar to that used in the external Hybrid Fibre Coax (HFC) network, and finally wireless. In turn, there are different standards for driving signals over these media, with the exceptions of Ethernet, and telephone wiring where HPNA is the only significant standard, from the HomePNA alliance.
Generally category 5 or 6 cable would be the preferred option because it has been perfected over many years for enterprise Ethernet networks, with interference problems well sorted out, enabling reliable provision of gigabit bandwidth. But it is costly and so only really viable for greenfields, mostly new buildings.
Powerline suffers most from fragmentation, with ETSI, IEEE and ITU-T all working to resolve the situation but themselves becoming bogged down in industry politics, leading to some complex hybrid proposals. The two main contending standards come from the Universal Powerline Association (UPA), and the HomePlug Powerline alliance. IEEE is trying to reconcile these options with its emerging 1901 standard, but this could mean having two different interfaces both at the physical level (PHY) and for data access to that medium (MAC). This will be counterproductive, according to Chano Gomez, marketing VP at DS2, a powerline equipment specialist based in Valencia, Spain. “DS2 believes that this gets in the way of improving the consumer experience. We are convinced that a standard should have a single PHY and a single MAC,” says Gomez.
Despite this, powerline has emerged as the leading contender for home distribution of video in much of Europe, where there are usually no other immediately available options, with little coax and patchy coverage of telephone wiring. In North America, the situation is very different, with many homes well cabled with coax, leading to that being the preferred option. The predominant standard is MOCA, but HomePNA is making an impact with a coaxial version of its standard. This allows operators to serve homes that have either coax or telephone wiring with a single home networking solution, just having to support the different PHYs within its gateway. However there is little interest either MOCA or HomePNA over coax in most of Europe.
That leaves wireless, which to date has been largely dismissed as an option for video, with the result that many operators are adopting a twin track approach – WiFi for broadband internet access, and one of the wired options for video, depending on the region and on what is already installed in a particular home. However, the case for WiFi has been strengthened with the arrival of the latest version, 802.11n. This doubles the transmission band from 20 MHz to 40 MHz and implements Multiple Input Multiple Output (MIMO) technology, exploiting the fact that radio signals bounce off objects and take multiple paths between source and destination, arriving at slightly staggered times. This was a problem, creating interference, but MIMO turns it into an opportunity, providing resilience, additional capacity, and improved coverage. It does not entirely eliminate black spots of coverage within a home, nor does it at a stroke deliver perfect, consistent QoS, but it is sufficient to enthuse some vendors and operators about the potential of wireless eventually to carry all communications within the home. Motorola, for example, is investing in video over WiFi with patented antennae designs, according to Eric Dowek, its marketing manager for Home & Networks Mobility.
Just to muddy the waters further, Motorola is among the vendors that have also developed a prototype plastic optical fibre, which is bendable with the aim of being installable by the user, capable of up to 800 Mbps. It has a much larger core diameter than traditional glass fibre, reducing its range considerably but fine for the short distances within the home. The main concern is over whether in practice it really can be installed by the customer, since it involves bending and some cutting.
FMC’s impact
Fixed/mobile convergence provides yet another twist to the physical network, with wireless operators promoting femtocell technology for the gateway. It is true that femtocells suffer from the same QoS issues as WiFi and are not being proposed immediately as an option for delivering broadcast TV to large screens. They are on the table though as an interface that gateways must support for delivering multimedia to mobile devices within the home, according to Paul Callahan, VP of business development at femtocell manufacturer Airvana.
“As femtocells become embedded into home gateways, the phone – for the first time – will be able to interact with the gateway, making the phone a personalised media device,” says Callahan. In this guise, femtocells offer one route to fixed mobile convergence within the home. It is the preferred route for many mobile operators because it is a natural extension of their cellular networks, with the gateway functioning as a small base station, extending coverage and capacity within the home. The alternative route is to leave the home gateway alone, and instead deliver fixed mobile convergence via dual mode handsets capable of accessing WiFi as well as the cellular networks as normal.
Whether or not they have to support femtocells, home gateways will have to cope with a plethora of interfaces. “There will be multiple home network solutions that coexist. There will not be one ‘winner’,” says Motorola’s Dowek.
Solving the QoS dilemma
Against this background it was almost inevitable that some initiative would emerge in order to crack heads together and create some form of order within the chaos. With the regular standards bodies dithering, the mantle has been grasped by the HomeGrid Forum, whose principal activity lies in the G.hn working group developing a common standard for interoperability among home networking devices such as gateways, irrespective of the underlying physical layer. In practice, G.hn will involve a new version of the three main wireline standards, powerline, phone line and coax in the first instance, but at any rate is broadly welcomed by most vendors. It defines a common technology for distribution of content over any wires inside the home,” says Ikanos’ Challa. “As a result one is able to develop a product that can be deployed on a world wide basis no matter which medium exists inside the home.”
G.hn then is an important ingredient for the architecture of the universal gateway, with other crucial components concerning remote management and QoS. These are beyond the scope of the HomeGrid Forum, with the key remote management standard coming from the DSL Forum with its TR69 specifications. This is fast becoming the de facto standard for managing home gateways and devices within the home network, having an established pedigree for dealing with DSL CPE equipment such as routers or modems. It has most of the right ingredients for gateway management already, with the ability to configure, activate and manage compliant devices within a home network from a central console, as well as update software, monitor status, and ensure that SLAs (service level agreements) are honoured.
This takes us to perhaps the most challenging aspect of all, QoS. This involves delivering packets at the average rate required for a given service, while keeping individual packet delays within specified limits that can be coped with by buffering, in order to avoid any transient artefacts. For this, the operator needs visibility within the home network, and the ability to prioritise traffic.
While remote management standards, such as TR69, provide the visibility, priority is then assigned by tagging individual packets within a stream, with favoured mechanisms including the IEEE 802.1p type of service flag, and Diffserv3. The difference is that 802.1p operates at layer two within the Ethernet frame, while Diffserv is a layer 3 mechanism within the IP packet. This means that 802.1p can only be used within a single LAN, since the flag is removed when it crosses a router, whereas Diffserv can be set remotely and operate at longer range across the internet or large IP networks comprising multiple routers. This makes 802.1p ideal for setting priorities within a home network, which is a single LAN. The flag has three bits, enabling eight different priority values to be set.
But it is one thing to set a priority and another to enforce it. For that reason, the home gateway will need the ability not just to set flags, but control all traffic within the network such that specific services can be given the bandwidth they require. This involves restricting bandwidth to other lower priority services, and also being able to liaise with the operator to ensure that each channel is given the right QoS. Such functions are provided by COS (Class of Service) mechanisms, which allow an operator to specify not just priority levels but also specific values for bit rate and other parameters for a given channel or service.
Given all these complexities, it is no wonder many operators are reluctant to trust home gateways yet to deliver full broadcast services. For the immediate future, many will persist with a twin track approach, delivering internet based services through the gateway, while confining broadcasting within home networks to trials. Full deployments will then come in 2009 or even later.
