Despite the temptation to talk about a single network underpinning the entire internet of things (IoT), in reality different applications will require different protocols. There has never been a single network technology underpinning the global internet, and there is no particular reason why there should be one for the IoT, if only because the standards process involved would take so many years to play out. Some ‘things’ will be connected to one another, and the internet, by wireline connections, some by wireless (in licensed or unlicensed spectrum); some will use the same established technologies as other devices, others may require specialized systems; some will run on public networks, some will demand the security and control of a private network.
While existing options like GSM and ZigBee have the advantage of being real and cheaply available, there is a sense that the industry should be developing new technologies too, designed from scratch to be optimized for various IoT scenarios. This is especially true in cellular, where LTE ‘ despite the fond hopes of operators ‘ was never created with vast numbers of ultra-low power connections in mind. It is too expensive and power-hungry to be optimal for most IoT applications, and its broadband capabilities will be overkill for most of them too, outside the fashionable areas of consumer wearables ‘ which are not where the main revenues of this sector lie.
However, these new developments should still not try to address every aspect of the IoT. The idea that one technology will work for the whole 50bn gadgets is neat but ludicrous. Most things will be communicating for specific and limited purposes that require a connection with certain characteristics ‘ ultra-low power perhaps, or dense meshing ability ‘ but do not require interoperating or roaming with all kinds of other IoT devices. The IoT will be global, but not a single network (any more than the current internet is).
It will embrace connected devices with different key priorities – massive broadband needs (future video systems), rapid mobility requirements (in-car platforms), ultra-low power (smart city or wearable safety sensors), very low cost. Some will only need to support low data rates, or occasional usage. And there will be short range links between the meshes of ‘things’, and long range connections to gateways and POPs and cloud servers. Just at the wireless physical layer there are Wi-Fi, LTE, GSM, Bluetooth, ZigBee, UNB, Weightless, 802.22, Z-wave and other emerging candidates with different strengths and weaknesses, not to mention proprietary approaches like the sensor mesh of Dust Networks (now part of Linear Technologies).
The cellular community generally recognizes that what it calls ‘cellular IoT’ will not be the whole picture, but it is desperate to ensure that LTE has at least some role in the business, in order to protect existing GSM-based revenues and allow cellcos and their equipment providers to generate new business from ‘things’. A group led by Vodafone and Verizon Wireless recently published a white paper discussing three main options for cellular IoT ‘ extending LTE, with the specific requirements of M2M in mind, but using existing bandwidth; doing the same but in narrower bands; starting with a ‘clean slate’ cellular standard. In all cases, they argue that in many scenarios, licensed spectrum will remain an advantage, for quality of service and global coverage (though some firms would argue that this should be specific spectrum, dedicated to IoT and not necessarily controlled by mobile carriers.
The white paper argues that it would be ideal to use existing cellular bands, sites and equipment and for a new standard to be deployable as an extension to 2G/3G/LTE. Clearly it would be ideal for current operators, since they could just add new services to improve the ROI of existing networks. However, the argument goes back to the debate about whether any aspects of current 3GPP systems are really suited to IoT, or whether the new networks are being shoehorned into these bands and technologies for the convenience of one community.
Elements which must be part of cellular IoT, in whatever form, are listed in the white paper and include deep coverage in ‘highly-shadowed locations such as basements, meter closets, manholes and even underground’, as well as support for low cost and sometimes disposable devices; ultra-low power consumption; optimization for small payloads to enable remote monitoring without substantial signaling burden; and an ‘adapted IPR licensing regime based on the FRAND principles and reflecting the reduced functionality that the new standard will provide for the specific M2M market (low cost, high volume)’.
The last point is a critical one, because it highlights an area where cellular norms are hopeless ill-suited to a world of huge numbers of very low cost gadgets ‘ its patent licensing. The prevalent system of bilateral deals, formed in secret between the scores of essential IPR holders in 3G and 4G, will be unworkable in the IoT, with its ultra-low BOM requirements and its huge variety of different endpoints. The unlicensed networking world is far better geared up for this particular aspect of the new networks.
The white paper gives a high level outline of what the clean slate option might look like, recommending a solution that would operate in a narrow 200KHz bandwidth using half-duplex FDD and single-carrier modulation with frequency and time division multiple access for downlink and uplink. The benefit would be that the design could be fully optimized for cellular IoT with no consideration for legacy functionality, and it could potentially operate in re-farmed GSM spectrum or in the guard bands of an LTE carrier.
Such discussions are interesting but still hint at an industry which is thinking in terms of one big standard, and one which has very different roots to that of the IoT anyway. As Jimmy Wind, CEO of the M2M division at Dutch telco KPN, told the recent M2M World Congress in London, M2M has been driven by verticals with their own networks and protocols, while mobile has been SIM-driven. A ‘one size fits all model’ no longer works, he insisted, arguing that to remain relevant, operators need to become technology agnostic. ‘Operators should look at what connectivity is available and extend the spectrum of connectivity that we provide to change the types of provision, whether it’s long range, small bandwidth etc,’ he said. ‘They need to be low power or else the business case will not be taken up.’