There are too many IoT protocols. Way too many. Anyone who says one single protocol will be the winner from end-to-end in all IoT applications and markets is smoking something. Software defined, multi-protocol gateways are the only hope on the IoT – and DSP cores enable this strategy.
A discussion I had a few years ago at CES provides the history we should understand. As you probably do, I have two boxes sitting just to my right that connect my broadband network: a cable modem (or DSL, same story), and a Wi-Fi access point and router. I decided to ask a stupid question of the consumer Wi-Fi AP vendors: why wasn’t there a single box, combo solution?
CES is always a crapshoot as far as finding someone knowledgeable on the floor who actually has time and inclination to talk. Dumb questions from a press badge are often blown off or dismissed with a canned response. After being fluffed in the PR sense by Linksys and others, I found a thoughtfully biased response from a product manager in the D-Link booth.
The first rationalization had to do with market differences. For example, DOCSIS is slightly modified between the US, Europe, and Japan, and at the time it was hard to cost-effectively build a single SKU meeting functional and qualification requirements for all regions. The second had to do with the rate of standards evolution. His thinking was with the continuing evolution of IEEE 802.16 and DOCSIS standards, one side of a combo box or the other would always be obsolete, discouraging buyers and increasing internal churn.
External churn in consumer markets is OK, in fact encouraged. Unspoken in that response is the consumer electronics model: reasonable prices, higher volumes, lower margins, limited repairs or field upgrades, and more frequent replacement sales.
This discussion was missing an important element, however: it predated broader acceptance of software defined radio with cost-effective DSP IP. At the time, DSP ASSPs or FPGAs were just too expensive for consideration in high volume consumer markets – and that hasn’t changed much. With lower cost, lower power DSP IP available such as the CEVA TeakLite-4, software defined radio strategies enter the IoT.
With many more protocols and versions in play, and longer lifecycles involved, the IoT makes a new case for a combo box. A software defined gateway can deal with protocols existing now, and new ones which may emerge as leaders in the future. An always-on DSP core provides hardware protocol acceleration to deal with complexities of modulation, extracting signals from noise, traffic shaping, and other necessities of a communications environment.
SGA Innovations has partnered with CEVA to provide such a software defined IoT stack, a combination of hardware and firmware IP built on a TeakLite-4 DSP. The intriguing part of this offering is it tackles both wireless and wireline communications:
- An 802.15.4 block implements a PHY/MAC supporting protocols including ZigBee, 6LoWPAN, and Thread. The solution includes an activity detector for wake-on-packet implementations.
- A powerline controller (PLC) block supports G3, PRIME, IEEE P1901.2, and HomePlug Green PHY. It can adapt to Cenelec A and B (Europe), FCC part 15 (North America), and ARIB (Japan) regulations, essential in supporting a world-ready design.
The SGA solution is synthesizable RTL that can be customized into an SoC. It moves the value-add equation from just a low-level block into a protocol-ready solution that protects IoT designs. (Had anyone outside of Google heard about Thread this time last year?) It also provides the performance to deal with multiple protocols simultaneously, essential for an IoT gateway.
Designers could certainly come up with a highly optimized IP solution for a single IoT protocol, but I expect to see many more of these multiple protocol, software defined IP solutions targeting the gateway space. Trying to anticipate where the IoT will go, and which protocol will be most popular, is wasted energy for most design teams. A DSP-based strategy hits integration, power, programmability, and acceleration targets – regardless of which IoT protocols emerge as winners.