When designing an M2M application, engineers will need to connect a cellular radio to the remote device. There are several considerations when deciding on which type of cellular radio to use. Each consideration basically involves trading off the cost of the cellular radio against other factors, such as performance, coverage, engineering effort and cost, and time-to-market.
The decisions to be made on selecting an M2M cellular radio are basically three-fold:
· Should you implement GSM cellular technology, CDMA technology, or both?
· Should you use 2G level technology or 3G (or maybe even 4G)?
· Should you use a separate modem, embed a pre-certified “socket modem”, or design in a basic cellular module?
Each of these trade-offs has unit cost implications on the final device. They also carry implications for available cellular coverage, cost of cellular coverage, product form factor, product longevity, and upfront engineering cost and risk. Each of these decisions can be made independently, but they should usually be made in roughly the order presented above. First select the cellular technology that will be used, then select the evolution of the technology to implement, and then finally select how deeply the cellular technology is to be embedded into the device.
Cellular Technology – GSM vs. CDMA
The first decision to make is whether to implement cellular technology based on the GSM standard or on the CDMA standard. This is a significant choice in the United States and Canada, since these are the regions of the world in which the CDMA standard is broadly commercially deployed. There are a number of new networks in other countries based on a CDMA technology called CDMA450, but it is not interoperable with the CDMA technology deployed in North America. If your solution needs to operate around the world, including North America, then implementing GSM is the only real option.
Solution developers have several factors to consider when deciding whether to implement GSM or CDMA. Both technologies have large coverage footprints in the US from two Tier 1 carriers each. Neither of these technologies have a marked coverage advantage over the other, but in less populated areas, there are places where CDMA coverage is available and GSM is not, and vice versa. The monthly recurring costs for service are roughly equivalent between the two technologies; in fact, there tends to be more difference between the rates charged between the largest and second largest carrier in each technology than there is between the technologies. There are also differences between the communications speeds or data rates provided by the technologies, with CDMA generally offering higher speeds in the basic, low cost technology evolution. M2M applications, however, neither require nor can take advantage of CDMA’s higher data rates.
One factor in which there is a clear difference between GSM and CDMA technology, however, is in the cost of the radio device hardware required to implement it. Regardless of the hardware platform selected – separate modem, embedded socket modem, or embedded module – CDMA technology costs at least 25% more than GSM technology. Table 1 shows illustrative prices for radio hardware for each of these two technologies in the summer of 2011 (in quantities of 1,000).
Table 1
Hardware Costs For Cellular Technologies
Cellular Technology
|
Separate Modem
|
Socket Modem
|
Embedded Module
|
GSM
|
$125-$165
|
$94
|
$23-$28
|
CDMA
|
$157-$167
|
$153
|
$30-$36
|
These cost comparisons are based on the actual selling prices of devices in the United States in the summer of 2011. The price ranges reflect the differences between various manufacturers, as well as the differences between different products with varying secondary characteristics (e.g., form factor, I/O count, brand, etc.).
Clearly GSM technology is the less expensive approach. Together with its global footprint, its lower cost has made it the most popular technology for M2M applications.
Although GSM technology may be the most cost effective choice, some applications desire to also have a CDMA design alternative to provide service in those locations where there is CDMA coverage and GSM coverage is unavailable. While this is easy to provide with external modems, and relatively straightforward for socket modems, the cost and complexity of providing supplementary CDMA support – in addition to GSM -- is significant when using embedded modules.
Technology Evolution – 2G vs. 3G
The second decision is whether to implement the most widely deployed cellular technology evolution – 2G – or to use the newer evolution that will soon become as widely deployed – 3G. Wireless carriers continue to upgrade the technology evolution of their networks is to provide greater capacity for handling cell phone subscribers with the spectrum they have available.
The original digital cellular data standard for GSM networks (called GPRS, not “1G”) is still available, as are modems and modules supporting it. GPRS offers no advantages over stepping up to the evolutionary level that is broadly deployed today – 2G. Carriers have recently upgraded most of their networks to the next evolution – 3G – which offers higher data rates and enables the carriers to support more devices on their networks. Carriers also are adopting 3G technology to provide better support for high bandwidth applications – sending and receiving pictures and video, playing on-line games, downloading applications, and making other file transfers. Most M2M applications do not need nor can they use the higher bandwidth provided by 3G technology.
Because 3G technology is just becoming available for M2M applications, it is primarily available through embedded modules. 3G options are not readily available for the other hardware platforms, socket modems and separate, stand-alone modems, although this should improve over the next year.
There is a significant cost difference between using 2G radio technology or 3G technology in an M2M device. Basic 2G embedded modules from major manufacturers can be readily obtained in moderate volumes for $26 to $30 apiece. On the other hand, 3G modules from major manufacturers still cost between $50 and $80 apiece, although those prices have come down significantly over the last year and are expected to continue declining. New 3G modules from less proven Chinese manufacturers are becoming available for as little as $38 at $40, at comparable volumes. Together with its lack of compelling performance advantages, the significantly higher cost has kept 3G from being adopted in M2M applications so far.
Carriers are expected to convert their networks exclusively to 3G some time within the economic life of many of the M2M devices being designed today. At that time, carriers are expected to discontinue support for 2G devices (at least with GSM). Consequently, and in spite of its cost premium, many device developers are beginning to move towards selecting 3G technology for their new designs.
Hardware Platform – External vs. Embedded
The final decision is whether to add cellular communications by using an external modem, or by embedding a wireless communications radio inside the actual device itself. External modems are self-contained devices in their own enclosures and often with their own, independent power supply. They usually connect to a standard communications port on the host device – usually serial, USB or Ethernet – and the device sends and receives data over that connection. On the other hand, embedded modules are designed to be built into the device, and they generally require a modification to the device design to support the module. Usually the module is attached to the device’s printed circuit board (PCB) and has signal and power connections with it.
In considering what type of radio to use, the basic trade-off is between the unit cost of the radio platform and the upfront, non-recurring engineering cost of supporting the radio.
With an external modem, the unit cost is relatively high because the modem is a complete, working device. It has its own power supply and enclosure. It also has its own embedded processor and memory, and the programming to drive it. There are also a PCB, connectors, LEDs and other components that add cost. On the other hand, to use an external modem, the device only needs to connect through a common communications port and implement some basic send and receive functions. The upfront engineering effort to implement an external modem is relatively low.
Embedding a wireless modem into a device, whether directly onto the main PCB or through an attached daughtercard, adds cellular communications capability to a device at the lowest unit cost. The only other major component required in addition to the module itself is some form of antenna (although even that can sometimes be implemented as traces on the PCB). On the other hand, the cost to build in a cellular module is significant. A PCB almost certainly has to be significantly redesigned to support a cellular module. Not only does space have to be made for the component itself, but the power supply usually needs to be modified to accommodate its high peak current draw. Since the cellular module is an active, relatively high powered RF device, the PCB also has to be designed to provide signal isolation and strengthen the grounding.
The other major upfront cost to embed a wireless module is the cost of certifying compliance with the different regulations governing cellular devices. The cost of testing by a certified lab can be as great as the engineering cost just to create the design. The engineering sophistication to achieve compliance is non-trivial, and is primarily a function of the product design. Engineers who are designing their first embedded cellular device will almost always fail to pass regulatory compliance on the first attempt. Redesign and retesting will add to the upfront cost of adding cellular connectivity.
“Socket” modems are an alternative to designing in the traditional embedded module. A socket modem still needs to fit into a connector on a main or host PCB, and there is some initial engineering effort to add that connector. But the socket modem avoids the need to conduct any additional certification testing, including for emissions, since it is entirely self-contained and pre-certified. The unit cost of a socket modem lies somewhere between the low unit cost of an embedded module and the higher unit cost of an external modem.
It is difficult to categorically estimate the upfront engineering costs to implement the three different cellular device alternatives because the design challenge differs so much from one product to another. One thing that can be compared, however, is the unit cost of the hardware. Those costs are evident in Table 1, above. The separate, external modem is the highest unit cost solution, while an embedded module yields the lowest unit cost. A socket modem provides some cost reduction over an external modem, but it is still significantly more expensive than an embedded module. In the case of GSM 2G, the socket modem is three times more expensive.
So which hardware platform should an application developer use? The answer must be determined on a case-by-case basis, and including all other considerations, such as form factor constraints and time-to-market objectives. But in general, when the application is going to be implemented in low volumes (such as for internal deployment within a company), then the most cost effective approach is to use an external modem. If the application is going to be deployed in any meaningful volume, of around 1000 to 2000 units (or more), then it is usually most cost effective to design in an embedded module and complete the regulatory certification process. The socket modem serves applications cost effectively that will be implemented in a modest volume. At a forecast volume of 1000 units, however, the embedded module can absorb almost $70,000 of incremental engineering development and certification cost (enough for many types of devices) and still be at parity with a socket modem.
Conclusion
Adding cellular wireless communications capability to a device requires careful consideration to select the most cost effective approach to using this technology. There is no answer that is optimal for all applications and situations. A good way to approach the problem is to first select the cellular technology, then select the appropriate evolution of that technology, and then to finally select the proper hardware platform of that cellular technology.
No comments:
Post a Comment