IPTV/VoD: solving the home wiring problem
The controls we need to install on all parts of the network are collectively known as QoS (Quality Of Service) or "traffic-shaping". These techniques are used very crudely by many ISPs right now to discipline chronic downloaders who manage to consume more than 100GB per month in data transfer. The underlying theory is the same – partition the connection into separate untouchable virtual 'channels' according to the application the data to be grouped is intended for. In a typical triple play setup, we use 3 channels which are cleverly prioritised, those being video (3Mbit/s), voice (276Kbit/s) and data (unspecified, or just the rest).
They are often not in real-time, expanding and contracting as and when necessary (a good example being video on-demand, as it only needs heavy bandwidth for the duration of a film, as opposed to live TV which is continuous). This partitioning can be done in multiple ways, but over an ATM network it is done using permanent virtual circuits (PVCs, or sometimes just called virtual circuits, VCs), and in IP-based environments such as LANs and MPLS networks, we use virtual LAN grouping (VLANs).
The trick for operators is how to get ATM and IP to play well together, which normally entails using a hybrid solution that maps ATM PVCs to IP VLANs. The ultimate goal is to move everything to IP by implementing a full IMS environment (PPPoE), which means most of UK DSL supplier community will have to wait for 21CN to be completed before they can migrate their own networks. Homechoice and KIT have both attempted to do almost everything they can as far away from a customer's home as possible. Both exclusively implement QoS in IP in their DSLAM and central office switches rather than cross streams with ATM, meaning the traffic that arrives in your home is already sorted and prioritised before it reaches your router and no filtering is required. DSLAMs are also crucial pieces in the delivery chain that suffer bottlenecking like no other, and are designed for aggregation.
There is another argument for exclusively using IP, which is derived from the concern that PVCs are simply not inherently scalable enough when you need a minimum of three per home. Further along down the line, we will need to partition further and possibly far more dynamically. Gamers have different requirements from office workers, as download fiends do to grannies just getting the hang on this 'interweb' thang. Quality of service engineering may lead to more tailored and high-performance lifestyle connectivity packages.
There is still much spirited debate amongst research scientists as to whether video can be better delivered at layers lower than IP across optical networks. Almost all networks use a multi-layer topology model called the Open Systems Interconnection Reference Model (or OCI model) to illustrate and describe how the transport traffic and function. The OCI model is composed of seven layers and is a very useful start when trying to understand the nature of transporting video over DSL.
Everything done on the backhaul network or in the DSLAM needs to be mirrored in the home network, which is very challenging to say the least. Once we prepare a standard QoS setup for our router/gateway to use, we need to consider how it administrates it between all the devices in the customer’s home. The dominant type of household that uses DSL is a family home, which despite the usual chaotic mess means that wiring has to respect the order of the nest. A customer's home can't have wiring all over the place, must be installed conveniently and be easy to maintain. Interior design is a key issue that almost no ISP seems to understand – it's even more foreign than the concept of 'customer service'.
There is an incredible opportunity for small local IT businesses to provide a call-out home wiring installation service on behalf of ISPs in the same way local TV servicemen and electricians work for Sky to install satellite equipment. It's not practical for big businesses with low margins to offer localised support when they need to consolidate their interests nationally.
Sending engineers to do installations is what most consumer industry pundits call "truck roll", and it's universally understood to be a very expensive activity that must be out-sourced or only used as a last resort. The biggest issue with self-install packages is not being able to deliver packages in the post as the recipient isn't present to sign for the goods that have been despatched (or CNP – 'customer not present'), and they are too big for the letterbox they need to go through.
Truck roll eats into profits like nothing else (especially as customers hate paying setup fees), but for the high-street it is unfortunately essential. The normal accountancy that pays for it is a relatively simple twiddling of the figures to amortise the immediate cost of the installation against the fees recouped across the subscriber's service lifespan. People like Sky have got this down to an art form as it is paid back in less than a year, but only the big names can afford it, as beside inflating customer acquisition costs, without a very understanding bank it kills cash flow dead. All the major brands use truck roll to alleviate as many support costs later on as they can – Sky, BT (broadband and the new Vision IPTV service), Homechoice, NTL, Telewest and more.
There are currently five choices for wiring up a customer's home so a TV signal can be carried between all the rooms: wireless/wi-fi, Ethernet/Cat-5/6 cabling, coaxial cables, home phoneline networking (HPNA – 128Mbit/s over telephone wiring) and Ethernet-over-powerline (e.g. HomePlug). All require additional cost and configuration of some kind. Soon after looking through all the options, it becomes apparent that we don't have the right tools in place to make IPTV the smooth, sweet and beautiful transition it could be. Most homes being built from scratch now directly wire optical cable to the front door and are channelling hollow pipes in walls that can accommodate wires of every variety. Importantly, there is nothing to say these techniques cannot be combined and mixed to achieve the same ends as favouring one standard mechanism.
Sponsored: Protecting mobile certificates