In Part II, the panel of experts from academia and industry address the opportunities or lack thereof afforded to RF designs in the IoT space.
By John Blyler, Editorial Director
Synopsis of compelling comments:
- “IoT itself doesn’t define an industry … but the verticals do.
- “IoT is at the top of the technology hype curve – at inflated expectations … But that is the perfect spot to do research … the point of maximum interest.”
- “Most consumer applications don’t really care that much about the RF performance … but it is critical for industrial applications.”
- “There are no earth-shattering areas of RF research associated with IoT/IoE It is all in how the upper layers work.”
- “It is too early to say who will benefit the most from RF research in IoT … but it will probably be the semiconductor companies.”
A panel of experts was convened at the recent IEEE International Microwave Symposium (IMS) to talk about the role and impact of RF subsystems in the design of emerging IoT devices. Part 1 of this series of reports focused on the new challenges faced by RF engineers in the IoT space.
Part 2 focuses on the IoT research opportunities created in the RF space. What follows is a portion of the actual panel discussion. – JB
- Ravi Subramanian, GM, Analog/RF/Mixed-Signal, Mentor Graphics, USA
- Conan Zhan, Deputy Director, RF Division, MediaTek, Taiwan
- Baher Haroun, Director, Embedded Processing Systems Labs, TI, USA
- Larry Larson, Dean of Engineering, Brown University, USA
- Gang Gary Xu, Director, Samsung R&D America, USA
- Gangadhar Burra, Senior Director of Technology, Qualcomm, USA
- [Moderator, standing on right] Oren Eliezer, EverSet Technologies & TallannQuest, USA
Part II – What Are the RFIC Research Opportunities created by the IoT?
Oren (Moderator): What research opportunities does the IoT create for us, specifically as RF people?
Ravi (Mentor): The research opportunities come from understanding the various vertical industries. IoT itself doesn’t define an industry. Instead, each vertical have specific needs and will drive specific research based upon what IoT/IoE means for that market. For example, the automotive, personal consumer health world or industrial markets have very specific research consortia and research topics that are getting funding. These are programs that are moving from per-commercial research into commercial deployment.
In terms of the hot topic, I don’t want to rain on everyone’s parade, but we all need to be aware of the technology hype cycle. As engineers and technologies, we need to digest this carefully because it tells us the story of every new hot thing. The cycle starts with a technology trigger to the peak of inflated expectations, the tough of disillusionment, the slope of enlighten to finally the plateau of productivity.
Oren (Moderator): Where are we now?
Ravi (Mentor): Great question. If you look at the latest Gartner slide (Figure 1), from March of this year (2015), IoT is at the top at inflated expectations.
Gangadhar (Qualcomm): For grad students, that is also a PhD cycle.
Ravi (Mentor): From a research perspective, there is a lot of funding and consortia. But as we ride down the tough of disillusionment, people will see these market size numbers shrinking as various verticals become real and others will struggle. Eventually, we’ll reach the slope of enlightenment where we’ll really see what IoT means. If you don’t believe these curves, look at the history of WiFi. Today, we take WiFi for granted, but it still has its limitations. But it has open huge opportunities in various markets. IoT will be similar. [Editor’s Note: Bluetooth is a classic hype cycle story.]
Conan (Media Tech): I don’t need to emphasis that low power is important. What can be expected? In 2030, expecting increase of radio per person. Regardless of whether devices are fixed or mobile, the density of RF-enabled devices will increase. From the radio perspective low power and duty cycle need to be taken into account.
- Today: 10 radios/person and 70B radios
- 2030: 100 radios/person and 800B radios ($100B+ Silicon opportunity)
Baher (TI): I’m going directly into what are the key research areas. It (IoT) may be a commodity on the consumer side, but look at the industrial applications with infrastructure and utility of costs, e.g., for monitoring oil/gas pipelines or their sensors. Pervasive monitoring will push the RF requirements much tighter. These markets might not be huge be they will be valuable and rich in sensor/RF applications. RF sensing trends include mmWave, THz sensing, and proximity detection as well as spectroscopy and triangulation/location.
I’m not talking about an IoT consumer, e.g., hanging a node on a dog to find the dog. Such consumer applications don’t really care that much about the RF performance. But to monitor an existing manufacturing line, you are limited to an industrial-based network like Profinet or Ethercat with 10/100 Gbit Ethernet card. You can not add another 50Mbs on top of that network without impacting data. So RF becomes augmented and complementive. It becomes a hyper-net. These are situations where latency, arrival and interference reduction are critical. The design must be very efficient on re-transmissions and guaranteed arrival time.
If you look at any RF-sensor system, there is a sequence of the energy flow (see Figure 2). There is the start-up, sensing and then sleep. You always do sensing first because that is what you want to transmit. You must sense the receive channel to figure out where you will transmit. The unknown is the “blue point” which is the sensor actuator (SA). This quantity can be all over the place and complete dominate your energy in the system. For the receive and transmit activities (red bars), the energy and cycle time should not be an issue. But the big start-up time tends to dominate your energy. Ideally, you’d like a fast start-up time where you are not spending a millisecond waiting for the crystal or synthesizer to wake up. These are problems that need to be solved so that you can do these short packets and not have a very low duty cycle.
Larson (Brown Univ.): I’d like to return to Ravi’s excellent, earlier “hype” slide (Gartner hype). The question is, are there great opportunities for RF research and to discover new things in the IoT space? According to this slide, we are actually in the perfect spot to do research. We are at the point of maximum interest in the IoT and maximum excitement about the process. Like all of these fields, it will head down the curve pretty soon. But in terms of the most fertile period for deep and impactful research, the peak of these curve is where you want to be. It is a wonderful time to be addressing this.
I want to mention a couple of things. First of all, I think the RF energy harvesting area is a really interesting one. We see more and more papers on this in recent years. It is an area where there is an immense amount of headroom to go before you reach fundamental limits. Typical RF energy harvesting now is in the 1 to 10% efficiency range. There is no reason why we can not do substantially better with improved devices and materials. There are some fundamental physical limits but we should be able to approach those pretty well.
Another hot topic is the merging sensors with RF in CMOS. I think that the capabilities of this IoT technology will depend upon how well we can integrate the sensor – field sensor, temperature, whatever – with the CMOS device.
My third hot topic would be creating a completely single chip 4g/5g radio with no off-chip passives for ultra-low form factor. It’s been a Holy Grail for many years. Plus, energy storage remains a chronic problem with a lot of these devices.
The final hot topic is CMOS compatible energy storage devices like ultra-capacitors, scaled batteries and the like. But batteries improve at a Moore’s Law pace but one that is 20 years for doubling (not 18 months). We need flexible batteries, low cost batteries. So, research topics that will address some of these fundamental RF need could really have a big impact.
Farshid (Samsung): Is there an opportunity for RF innovation? Sure, but you really have to rank the IoT opportunities. I rank the key IOT challenges that are also opportunities as follows:
- Interoperable systems and standards
- Open access wireless protocols and custom protocols
- Multiple standards – additional power and security overhead
- Security challenges
- Security and low power requirements – contradictory
- Expensive traditional security techniques
- New research directions – physically unclonable functions (PUFs), and optimized AES architectures.
- Ultra-Low Power Sensor Nodes
- Adapting existing ULP techniques like sub-Vth, Near-Vth, adaptive clocking, etc.
- Leakage at advanced technology nodes reduces retentive memory density
- Solid state batter densities and energy harvesting modalities.
Add to this list new types of sensors perhaps in the health or other areas. There will be many challenges at the integration, architectural and security levels. So I still stay on the low side on how much is there for an RF engineer, if you are really limited to the RF. I’m not certain whether somebody can actually graduate at the PhD level just doing the radio part of this connectivity challenges. How does it differentiate from existing research in low power Bluetooth (BT), FSA transmitter or existing types of radios? But there will be lots of work in optimizing the entire solution.
Gangadhar (Qualcomm): The word Internet of Everything (IoE) is loosely used in our business. Baher (TI) is probably right in noting that you can consider things like geo-physical sensing or some reasonably complicated, esoteric application as IoE. But, for the most part, that is not what comes to mind … you know, it is one of those word games: “tell me the reaction when you say me the word IoE.” I guarantee you that geo-physical sensing or something like that is not it. So, I’m not disputing what he said, just that is the perception.
In my opinion, if we define IoE somewhat inaccurately. Is it a set of devices for low power, low cost applications? Or are these devices applicable across a wider gamut that includes cloud connectivity, automotive and the like? If so, then from an RF perspective, there is not a whole lot of research needed. I’m sorry, but for RF design, there are no earth-shattering areas of RF research associated with IoE as we define IoE. I go back to my previous comment. It is all in how the upper layers work. It is all in how the complete system works. That’s the way I see it.
With that said, what do we have as potential (research areas?). There are still some corner cases or still important elements. How do you do multi-function radios, i.e., adaptive radios for low cost IoE? How do you do a single piece of Silicon that can adapt itself across different standards? Farhair talked again about interference robustness. I totally agree. That is a huge element that must be taken into account.
Another area of opportunity is energy management. How do you do the overall power/energy management in relation to what those circuits drive as the load current from an RF component or an non-RF component in an IoT system. Those IMO are some of the areas some work could flourish.
Oren (Moderator): Let’s see what the audience thinks. Who is really going to benefit the most in the area of RF research in the IoT market? According to the polling (see Figure 3), most folks in the audience believe it is too early to say but most probably the semiconductor companies will receive the most benefit.
Part 3 of this panel discussion will address which of the many competing consortia will eventually dominate the IoT RF market.