IoT implies massive number of devices, low data rate and non-real-time, and no connected mode mobility. So these “use case” conditions are being applied to both LTE and 5G to enable higher capacity and cost efficient IoT networks. We will look at some of the specific technical issues and solutions raised by this.
LTE has a signalling overhead to maintain many live connections (Polling processes) so that all devices registered in the network can be contacted at any time, and all devices can request to send data at any time. This process has a limited capacity, and takes capacity from the available “user data” bandwidth to use it for this “control data”. In addition, the “mobility management” function ensures that the device is always connected to the best available network cell, and performs a number of measurements and reports to evaluate available cells (including other networks, other frequencies) to select and re-connect to the best service. This is critical during a “handover” where a connection (e.g. voice call) needs to be maintained as the device/user moves between coverage of different cells. All of this mobility management function requires additional signalling capacity, and additional functionality and capability on the user device.
This presence of mobility mechanisms and constant connectivity, versus intermittent data and non-real-time transmission, is a key differentiator for IoT networks. The IoT network has no ‘connected mobility’, only ‘cell re-selection’. This means that if the IoT device moves from one coverage area to another, then a re-connection time is needed before the device is re-operational on the network. This may be a few seconds, or may be many minutes of delay. For IoT data transmission this is acceptable (for voice calls and internet browsing this is not acceptable).
As we look forward to new IoT waveforms (such as NB-IoT) and to 5G we see key changes in concepts, to remove the overheads inherited from voice/browsing use cases and to introduce more capability in the access methods. These are described in the following sections.
‘Connected mode’ is redefined, as we no longer have a use case where on-going and continuous transmission/reception of data is required. So we remove the need for signalling and resource management to activate the transition from ‘idle mode’ to ‘connected mode’, and instead the IoT data is simply sent over a discrete burst of traffic. If the burst is sent at the same time as another burst from another device, and is not received by the network, then the device simply re-transmits after waiting a short period of time, and this is repeated until finally the data is sent successfully. This “un-coordinated” approach does introduce extra latency and re-transmissions, but removes a complete layer of complexity in terms of signalling and scheduling of data, allowing for more users onto the network.
Mobility is reduced, as we no longer have the use case of a device transmitting continuously whilst moving from one cell coverage are to another. This effectively reduces the need for measurements and reports from the device (to inform the network of coverage from suitable neighbour cells), and hence reduces the extra signalling and frees up extra capacity for data bursts. This also reduces cost (simplicity) to implement the IoT device modems.
DTX/DRX sleep modes are extended for longer periods for longer battery life. This reduces the volume of signalling in the network, as most devices are in “sleep mode” for long periods and hence do not require any signalling capacity to be reserved for their possible use. This means that many more devices can be managed within the cell, as it is assumed (or controlled) that a large majority of them are in pre-scheduled sleep periods and hence require no signalling/processing capacity. So we can see an extra benefit of this feature, used for longer battery life, is to expand the support for high numbers of devices.
Finally, in 5G we expect to see spectrum usage and access methods redefined, e.g. new waveforms like NOMA and SCMA, to support more device capacity. These enhanced access methods introduce an extra level of resource allocation/access on top of the OFDMA used by LTE. NOMA uses the power domain, to multiplex near and far devices onto the same frequency resources, and uses Successive Interference Cancellation (SIC) to then separate the near and far signals at the receivers. This effectively doubles the capacity/number of devices, as many more users share the same frequency resources, assuming they are spread in distance from the cell centre. SCMA uses code domain multiplexing (used in 3G) but combines this with OFDMA (used in 4G LTE) to give an access method that can provide much higher capacity than either 3G or 4G can. SCMA shows high possibility to support many more simultaneous transmissions within the cell, and could provide for very high capacity in terms of numbers of separate users all active within the cell.
In summary, we can see that there are significant changes in the amount of signalling and connectivity required by IoT use cases, and hence LTE is adapting to this and creating new variants optimised for the high numbers of users envisaged for IoT. In addition, 5G is investigating new access waveforms that can give significantly higher gains in user capacity by using new access methods such as NOMA and SCMA.