The amount of white space is typically proportional to the total amount of TV spectrum – roughly halving the TV spectrum results in half the amount of white space. In some cases, the reduction may be slightly greater if a TV repacking process results in tighter use of TV frequencies and so less white space, but typically in Europe TV is already packed as tightly as engineers can manage. So any reduction in TV frequencies will reduce white space.
Current calculations of white space availability are generally based on the UK. Here, there has been an assumption that as well as the 800MHz part being used for cellular, the 600MHz part (from 550-614MHz) would also be auctioned and hence not available for white space. So the remaining UK spectrum for TV transmissions is 256MHz. Calculations suggest an average of around 100MHz of this is available as white space in any location.
If, instead, the UK decided to adopt the WRC plan then the TV transmissions would be 470-694MHz – 224MHz (perhaps less 8 or 16MHz for channels reserved for radio astronomy and wireless microphones). Scaling for the drop from 256MHz to 224MHz we might expect white space availability to fall from 100MHz to 88MHz. Such a small decrease is unlikely to make any material difference to white space usage.

Hence, an initial conclusion is that were countries to follow the WRC recommendation, the amount of white space left would not be materially different from that which most have been using in their calculations to date.

It is worth noting that the position may be better than this. Firstly, many European countries will find it very difficult to open up the 700MHz band without at best massive re-planning of TV transmission and at worst reducing the number of TV channels transmitted. This will likely mean that not all adopt the plan and many do not do so until well after 2015. Secondly, if 700MHz is replanned for cellular it is likely to be on the basis of frequency division duplex (FDD) with an uplink and downlink and central guard band. It is quite likely that this guard band would be available for white space usage and would form a set of national channels on which there was no licensed use, making them “superior” white space. Not only would this add back enough channels to bring the available white space back to around 100MHz it would actually increase the quality of the spectrum available. Thirdly, restricting the bandwidth of white space will allow more efficient antennas, increasing the efficiency of usage slightly.

So in summary, the WRC decision is not likely to have any impact for some time, at worse case results in a small reduction in white space availability, but may actually improve white space access.

Weightless has a similar structure to a cellular system, having a core network with many of the same functions. But rather than needing big machines it will run entirely in the Cloud. This might seem rather incredible when a machine network is likely to have many more subscribers than a cellular network. But a machine network has a number of simplifications over a cellular network including:

• All traffic is genuinely packet based. Hence a fully IP-based switching solution can be used without any need to support time-critical applications such as voice over IP.
• A Weightless network will typically only have around ¼ of the number of base stations of a cellular network.
• Seamless handover does not need to be supported.
• With longer latency than is typical for cellular systems there is much more time to process data allowing for lower performance platforms.
• Cellular systems support a wide range of services such as voicemail, short code dialling and much more that is not required for machines.
• Overall traffic volumes are much smaller. While a 4G base station might generate 100Mbits/s of traffic or more, a Weightless base station will typically generate 100kbits/s on average – some three orders of magnitude less.
• Cloud-based computing platforms are becoming rapidly more powerful. It might be that a re-design of a cellular network would result in some of the core functions being moved to the cloud.

Cloud-based hosting of the core network is extremely advantageous. It means there is no initial outlay for the core network, the processing power can grow as needed and backup functionality / quality of service can be provided by the Cloud supplier rather than through deploying redundant solutions. Upgrading to new functionality is simple. It may appear a rather radical idea to those used to buying, owning and operating their own infrastructure, but equally it could be seen as just another step along the lines of outsourcing the operations and maintenance of a network.

In the case of Weightless it means that the minimum outlay needed to become an operator is close to zero, scaling with coverage and usage and that the development of the core network can proceed much more quickly than would otherwise be the case. It opens the door to a wide range of new business models and allows new types of interaction between different networks.

One day all networks will be built this way….

But the likelihood of a successful transmission is not just whether the operator has spectrum. There has to be coverage and capacity as well. Many will have had the experience of not being able to make a mobile call either because of a lack of coverage or due to “network busy” effects. Capacity itself is linked to spectrum availability but includes other factors such as number of cells and policies around sharing resources. The experience of many is that their home Wi-Fi, using unlicensed spectrum, is typically more reliable than cellular networks. And Wi-Fi networks are increasingly being built – and relied upon by cellular operators for data offload from their networks. So there is much reason to challenge the orthodoxy that reliable communications requires licensed spectrum.

So what about white space? It provides excellent propagation, so coverage-related problems should be fewer. There is also, on average, a large amount of available spectrum – of the order of 120MHz. This is equivalent to the entire 3G band, shared among typically four cellular operators. If there were no interference from other users this should result in reliability much greater than can typically be achieved with a cellular network. The imponderable is how much interference will occur. This depends on how many other users try to access white space, how powerful their transmissions are, what technologies they use and whether there is any coordination amongst users. Simplistically, if there were four systems trying to access the spectrum then this would lead on average to each having the same amount of spectrum as a 3G operator and hence reliability might be expected to be similar. But there is much more to the reliability question than this.

System design is a key issue. There is much that can be done to make a system more reliable. At the radio level techniques such as frequency hopping and dynamically variable spreading can ameliorate the impact of interference. At the resourcing level, resources can be reserved or prioritised towards more important messages such that congestion does not impact high-priority signals. Weightless makes use of all these and more.

Co-planning can also be valuable. Where the other users of the spectrum also have networks, then it makes much sense for the operators to work together to share the spectrum in a way that minimises the interference they cause each other – cellular operators are familiar with doing this in border regions. It might also be expected that as the spectrum became more heavily used it would dissuade new operators from building networks because the risk of interference-related problems would grow. Hence, there would not be a “tragedy of the commons” where the spectrum became so congested that nobody could sensibly use it. This is less effective where the other users are local, such as individuals running home Wi-Fi systems, although the impact of their interference is correspondingly more localised. Actually, we currently do not expect non-networked systems to use white space because of the overhead associated with the “geolocation” process which adds cost to these systems.

Ultimately there is no certainty in wireless transmissions. Cellular systems can be congested, suffer from malicious interference or have equipment failure. White space systems may have less certainty concerning interference but more spectrum leading to less congestion. It is possible to envisage a scenario in a geography with little white space available where multiple networks result in serious interference and even prioritised messages suffer. Our assessment of the likelihood of this happening is that it is no worse than congestion occurring on cellular networks. But there are so many unknowns that it will never be possible to be definitive about reliability on any wireless network.

If absolute reliability is essential use a wire. If a wire is not practical then wireless, even without absolute reliability, is likely better than nothing at all.

The wireless machine-to-machine (M2M) market exhibits just these attributes. The cellular industry is obsessed with ever faster data rates, ever greater capacity (and ever more complex and expensive devices). The 4G standard is the epitomy of this. For everyone in the industry this has been a successful and profitable obsession and with smart phone use growing it looks like it may continue in this manner for a while yet.

So a new technology that has a lower data rate and less functionality does not really compute for the cellular industry. They tell those in the M2M world that their technology can be adapted to meet requirements, that cellular is well established and that perhaps a variant of 4G can be produced that is cheaper but slower. None of this is actually wrong, but if you wanted to design a Toyota you probably wouldn’t start with a Ferrari and cost-reduce it.

Weightless is a technology designed specifically for M2M. Compared to cellular technologies it is slower and lower capacity, but crucially it is an order of magnitude cheaper and meets requirements such as 10-year battery life. Just like the smaller disk drive in Christensen’s example, it does not follow the previous trends – it truly is disruptive. And it does lead to a dilemma for the mobile operators: should they try to support this requirement with their cellular technology or is it time to embrace an additional “product line”?

The Innovator’s Dilemma shows us how difficult it can be for industry leaders to respond to disruptive technologies. It also shows what happens when they fail to do so.

There could be a simpler solution. The home router could be configured to send information such as passwords to a trusted site in the Cloud, perhaps linked to a credit card security system. When a purchase of a connected device was made with the credit card (and properly authenticated) then details of the device (eg an IPv6) address could be passed to the trusted site. When the device – perhaps a new TV – was turned on in the home it could initially use an M2M communication link to talk to the manufacturer’s or retailer’s network from where it would be re-directed to the user’s trusted site. On authenticating its IPv6 address the site would then provide it with details of the user’s home Wi-Fi configuration. The TV could then initialise its Wi-Fi connectivity and link into the home network. The M2M link might never be used again, or only in the cases where Wi-Fi connectivity failed in order to determine whether new configuration details were needed. For the user there would appear to be an almost magical “plug and play” connectivity of new devices into their home network.

It might seem rather excessive to build in a communications system that was only used once by a device, but if the costs of the chipset were low enough and there was no on-going burden on the M2M network from the device remaining passive then there seems little reason not to do this. Indeed, the cost savings on helpdesks and returned devices might more than justify the chipset cost.

Such an approach would require an amount of standardisation work. The Wi-Fi community is working on a solution to this problem for roaming devices within the HotSpot standard and it is likely this could be leveraged to enable home routers to place their credentials into a managed site. M2M technologies capable of meetings these requirements and cost points are well on their way to deployment. The missing piece of the puzzle is a way to link the device just purchased with the remote Wi-Fi credentials in a secure manner – perhaps something a credit card company might like to take forwards as a value added service.

An approach often adopted by regulators to overcome this complexity is to take the worst case – often called the “minimum coupling loss”. This assumes the smallest possible separation between devices, the worst case alignment of antennas, a complete lack of any obstructions, the worst case receiver and more. This does ensure no interference but at the expense of excessive regulation. For the worst case the loss of signal between white space transmitter and TV receiver is around 43dB (the bigger the loss the better as it means white space device transmissions are weaker when they reach the licensed receiver). However, some modelling suggests that the probability of this occurring is infinitesimally small. Indeed, the probability of a loss of less than 53dB, 10dB above the worst case, is less than 0.1%. If the worst case were taken as 53dB rather than 43dB this would more than double the range that could be achieved from a white space device, making the service it can provide much more valuable.

So what should be done? Should regulators play absolutely safe and insist on 43dB minimum coupling loss. Or should they aim for a non-zero but very very small interference probability that would massively increase the value of white space spectrum. The answer to that question typically depends on whether you are a licence holder or someone who might benefit from white space usage…..

All the experts would agree that a mix of solutions is needed, coupled with adequate subsidy. Wired solutions, using fibre optic cable or DSL over existing copper pairs works well in denser areas where homes are close together and the cost of digging up the road is relatively low. Outside of these areas wireless systems tend to offer better economics, while in the most remote areas satellite links may be the only viable solution. For the most part, rural communities are too isolated for wired solutions to be viable, even with subsidy. So the focus turns to wireless. Any wireless system has to balance coverage and capacity – generally the two are incompatible. Coverage is best delivered using lower frequency radio spectrum where signals “propagate” further. Capacity requires large amounts of radio spectrum which are generally only available at higher frequencies. This trade-off is broadly the reason why high speed mobile services are typically less available than lower speed “2G” mobile coverage. Unfortunately, rural broadband requires both coverage and capacity – offering a broadband service below 1Mbits/s is not, by definition, broadband. So wireless has not to date made any real impact into rural broadband.

This is changing due to the availability of some radio spectrum that is both at low frequency and plentiful. This is the so-called “white space” spectrum interspersed among the TV bands. It propagates well (most can get TV transmissions) and in rural areas there is typically more spectrum than the entire 3G cellular spectrum. This spectrum is shortly to become available in the UK thanks to some far-sighted regulation by Ofcom. Even better it is free, reducing the subsidy needed and allowing local communities to deploy their own systems rather than having to persuade national operators to show an interest in local problems.

Perhaps it is a good thing Councils have been unable to progress rural broadband to date. They might otherwise have selected a solution before the new white space systems were available. They do now need some expert help to understand the options open to them. Fortunately there is no shortage of wireless experts in the UK. Even more fortunately, white space technology is available from pioneering UK companies. I work for one.

However, in the case of wide-area machine communications, there are no standards to choose from. Technologies like Bluetooth and Zigbee are for local communications, not over many km. Technologies like cellular are for wide-range but for people, not machines. While they can be adapted to machine use they are expensive, inefficient and unable to meet requirements such as decade-long battery life. Someone wanting to implement machine communications for, eg automotive engine management, at present does not have any standard to choose from. But standards are critical when multiple different applications need to share the same access networks and roaming globally may be required. Indeed, history has shown that wireless communications systems do not succeed unless they are standards and further that in most cases there is only room for one, or at most two global standards – examples include Bluetooth, WiFi, cellular, digital TV standards and more.

The aim of Weightless is to meet this need. Of course, others have sought to do so in the past but not been able to make progress due to a lack of radio spectrum that is affordable, globally harmonised, plentiful and in frequency bands with long-range propagation capabilities. What has changed this is the availability of “white space” spectrum in the TV bands which meets all of these requirements. But white space spectrum comes with its challenges including potential interference, both from the licensed users such as TV transmitters and other unlicensed users, a need to work with geolocation databases and more. These require quite specific characteristics that are not available in any existing standard – existing standards tend to be either short range and operate in unlicensed spectrum or long range and operate in licensed spectrum. So using white space enables machine communications but provides additional reasons why a new standard is needed.

Some of the key characteristics needed from a machine standard include optimisation for short messages (often less than 50 bytes), ability to handle up to 1 million terminals per cell, support for ten-year battery life, excellent authentication and encryption, global roaming, deep indoor penetration, tightly constrained radio emissions to avoid interfering with others and the ability to work in a environment where interference is uncertain.

The process of developing standards is not always a smooth one. Conflicting company priorities and the desire to secure essential IPR can delay and sometimes even derail a standards process. That is why Weightless is developing its own standards body – the Weightless Special Interest Group (SIG) and why it is insisting on royalty free licensing for terminal devices.

So happily, in this case, the nice thing about a machine communications standard is that there may only will be one to choose from.