It seems many are convinced that the magic number for device cost in cellular M2M applications is $5. What should be the cost of a module, and how critical an issue is the device cost to start with, is a topic of debate and although I have my view on it, it is not what I will address here*. Instead, I like to focus on what needs to happen to get the cost to sub $5. So, how can we reach such a target price and what are the implications?
Cellular M2M (or cellular IoT, we will use the two terms interchangeably) includes connectivity of machines through the cellular network, be it 2G, 3G or 4G. Today, 77% of connections are using GSM/GPRS where the cost of modem for M2M applications is estimated at $10. Modems based on 3G and LTE are priced much higher, and unfortunately, are not experiencing the sharp drop in price as GPRS modules have seen recently. In fact, much of the 3G and LTE solutions use the same silicon ecosystem as what goes into the phone, so prices remain high (>$50 for LTE module).
GSM/GPRS Module.
There are different factors that drive the cost of a module but two fundamental issues are critical: integration and volume. Volume is a perennial question of a chicken and egg. In our analysis for low cost cellular M2M silicon, several tens of millions of devices is required for a vendor to breakeven and a volume cost breakdown in the hundreds of millions before any substantial cost reduction is realized. Obviously this requires standardization of technology which justifies the flurry of recent activities at the standard bodies. Nevertheless, volumes will need to come from a major application whereby cellular M2M is rolled out in a ‘cookie-cutter’ simple process. What can this application be?
The other track, integration, also plays a pivotal role in reducing cost. There are many layers to address in this regard. Our observations on how cellular M2M device integration may proceeds points to integration of both baseband and RF function in one silicon device. In this case, it helps to limit the number of frequency bands to a minimum, for example, one band. This requires consensus to standardize M2M service on select few spectrum bands where economies of scale can be reached. Moreover, having a single band operation simplifies the RF front end significantly and reduces cost.
Another aspect is to reduce complexity of the technology which lowers the cost of silicon integration. Reducing complexity has many aspects that include reducing the features and requirements (ironically many M2M modules used today support voice service which is not required in machine applications!). The impact can be at different levels including lower memory requirements and less computational complexity that contributes to saving power. Reducing complexity is a key issue in cellular IoT where a number of proposals are vying for adoption as standards.
To enable greater integration and reduce complexity, we observe a push to reduce transmit power which helps to increase battery life – a key feature of IoT deployments. This is tricky because it can compromise range. It is envisioned that spectrum for M2M applications would be in the sub 1 GHz bands. While propagation characteristics in these bands are favorable for long range propagation, the antenna efficiency would be low for small modules whose physical size would not allow high gain. Lowering the transmit power is a proposed feature of LTE-M in Release 13, but it would be coupled with capabilities to increase the repetition of transmissions to claw back on link robustness and coverage range. On this note, LTE Release 13 calls for reduction of the number of antennas to reduce complexity and cost (I presume SISO will become a very popular term now!).
Can cost low cost ($5) cellular IoT devices hit the market quickly? Not before late 2017 or early 2018, but even then, hitting the actual cost target will require ramping up the volume. This is bound to make room for innovative solutions provided an available use case that can drive deployments. From this perspective we say to mobile network operators to be careful on what’s coming next!
* The cost of the device is one aspect, the other is the cost of service (and the type of service). This is what makes IoT such an interesting topic!
Frank, interesting read! M2M modules may always be at a higher cost in part due to the module maker value add and profits. There’s definitely value added to the silicon by the module assembly, certifications and firmware toolbox. The radio module itself is often considered a cost benefit up to 5K or 10K EAU while chip down is cheaper thereafter, but this doesn’t really apply to M2M since I don’t believe the cellular transceiver chip vendors want to support small customers without a huge cash outlay for support. Also the fine BGA or LGA pitch of those devices drives the bare PCB to higher cost technology. Perhaps the burden is on the chip vendors to help us get to $5, but the business case may not be there. Could innovative packaging of the bare silicon get us there? And what about the limited 4G bandwidth. It may be self serving of the chip vendors to preferentially support phone and tablet manufacturers as those customers sell more BW to fewer clients in contrast to M2M sensors. BW is finite so feed the lions and let the scavengers get what’s left over.