Presentation by Stephen Pattisson, VP Public Affairs, ARM. Technology Ventures IV, 19th October 2012, Manchester Metropolitan University Business School
3. ARM Partnership Momentum
Millions of units shipped per quarter
2,500 2002: 1 Billion
2,000 Cumulative
1,500
1,000
2010: 25+ Billion
500
Cumulative
-
1Q05 1Q06 1Q07 1Q08 1Q09 1Q10 1Q11
2020: 150+ Billion
Cumulative
3
4. Over 4 Billion people connected
via mobile phones
Smartphones will
leapfrog over the PC
in the developing
world
Smartphone data
traffic will exceed PC
traffic in 2014
Over 95% of all
mobile phones have
an ARM core
BIG NUMBERS!
4
5. It’s Not Just the Apps Processor
Bluetooth
Cortex™-M3 & Cortex-M0
WiFi
Cortex-R4 & ARM968™
Bluetooth WiFi
Cortex-M3 & Cortex-M0 Cortex-R4 Cellular Modem
Cortex-R4, Cortex-R5 &
SIM Cellular Modem
Cortex- R7
Securcore® SC300™ Cortex-R4, R5 & R7
SIM Flash Controller
SC300 Cortex-M3
Flash Controller
GPS Cortex-M3
Cortex-M3 & Cortex-M0 Apps Processor
GPS Apps Processor
Cortex-A5 (MPCore),
Cortex-M3 & Cortex-M0 Cortex-A5 (MPCore),
Cortex-A9 (MPCore)
Cortex-A15 (MPCore)
Cortex-A9 (MPCore);
Power Management
Cortex-M3 & Cortex-M0
Power Management
Cortex-A15 (MPCore);
Cortex-M4
ARM Mali™ Graphics
Cortex-M4 System Cntrl
Cortex-M3 & Cortex-M0
Mali Graphics
PLUS PLUS
CoreLink™ AMBA IP
System Fabric Touchscreen Controller
Touchscreen Controller
Artisan ® Physical IP 65-28nm
Physical IP 65-20nm Cortex-M0
Cortex-M0
Development tools
RealView Development tools
5
6. Smartphone Innovation Accelerating With ARM
800
100
2x Cortex-A15 700
90 1.5GHz
80 Cortex-A15 combined with
600
LTE will deliver a new
Smartphone Volume (Mu)
mobile experience
Relative Performance
70
4x Cortex-A9 500
Multicore Cortex-A9 1GHz
60
delivering performance
and low power 400
50 Developers start investing
heavily into mobile apps Cortex-A8 combines with 2x Cortex-A9
3G delivers a completely 1GHz
40 new mobile experience 300
Major investment into
30 mobile S/W from Cortex-A8
Google, Adobe 1GHz 200
Cortex-A8
20 ARM 11 starts the 650MHz
Smartphone Revolution ARM 11
ARM 926
100
10 500MHz
ARM 926 200MHz
100 MHz Source: Gartner Q4 2010 Forecast
0 0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
6
Back in 2002, ARM’s partners passed the 1 billion cores shipped mark – some 12 years since the birth of ARM. It only took an additional 8 years to pass the 25 billion core milestone and we are predicting that by 2020, the ARM partnership will have shipped over 150 billion cores – which is a BIG number however you look at it. By volume, the ARM architecture is the number one architecture in the world and with the help of our partners, we have a goal to become the number one architecture by value.
Those that have heard of ARM, mostly think of the mobile phone. On this slide there are some very interesting statistics that have been discussed at length in the press.Over 4 billion people now have a mobile phone and it is predicted that most of the developing world will skip the era of desktop or even laptop computing and use a smartphone as their primary device.They are also predicting that smartphone data traffic will exceed that of PC traffic in 2014 which demonstrates the world’s reliance on truly mobile computing. What you may not already know is that over 95% of all mobile phones have at least 1 ARM core with many having more than one
In many smartphones there can be 6 or more ARM cores. The most well known is the application processor which can range from the Cortex-A5 for budget smartphones up to the most recent Cortex-A9 and its big brother, the Cortex-A15 that is about to appear in the market. However, it is highly likely that there is a Cortex-R4 in the baseband module and a Cortex-M3 in the Bluetooth or Flash controller. GPS modules, WiFi and touchscreen controllers are also often ARM along with the power management chip. Even the SIM card could be built on a core from our Securecore family.So why am I presenting to an audience interested in M2M about the insides of a mobile phone. Well, it is because it is the mobile phone that is driving innovation in performance per watt.....or i should say, per milliwatt!
increasing demands of the mobile phone industry - growing adoption of smartphones - consumer expectation is producing ever more powerful applications processors. apps processors need companion chips that are just as power efficient so that a reasonable battery life can be expected from the latest and greatest smartphonephase of super innovation and growth driven by ARM’s accelerated roadmap - silicon partners investing heavily in innovative new SoC’shigh performance SoCs are becoming increasingly applicable to a whole host of additional applications and SiPs are now actively promoting industry specific variants - enable a whole new breed of low cost and low power platforms ideally suited to the needs of the M2M and sensor network developers
Some believe IoT to be the growth in RFID tags that identify every object to something else when placed in close proximity. This is one view but it is only a fraction of what IoT can achieve. RFID tags can be thought of as disposable information carriers best suited for one way communication such as a ticket or a clothes label. They may be embedded in billboards or used to check assets in and out of locations but will require coming close to a reader in order to become active.
Some people talk about machine-to-machine or M2M where devices communicate with each other without human intervention. This gets closer in my view to what IoT is which is why I often interchange the two but in reality IoT to me is where every object has a digital identifier that is either shared passively, as in the RFID case, or it communicates information that it knows about itself or its surroundings to something else – that something else being as simple as a data aggregator or it could be a data service on which other services are built, such as energy management or irrigation schedules.Cortex-A based platforms a particularly suited to the embedded computing space. screen based HMI - case of Industrial Automation or a car Infotainment system. powerful SoCs with a Memory Management Unit or MMU that enables a full feature or high level OS to be run. What about all the other devices in the system? - What about the medical monitoring devices in a personal health system? What about wireless sensors that form part of an energy management system?other end of the spectrum, what happens to all the data that gets collected? How is it manipulated? Do we do something with it before we send it to the cloud?This is where ARM scales. From the smallest sensor to embedded servers. And the concept of connecting everything together, we call the Internet of Things
For me, the internet of things is where connected devices and sensors provide regular or real time data to an embedded platform that then takes some appropriate action or presents that data in a meaningful way, either locally or through a cloud based service.As an example, on the slide we can see a child’s toothbrush communicating with an information display built into a bathroom mirror that is also pulling data from other sources. The mirror contains a camera and uses facial recognition to present data pertinent to that individual such as messages, itinerary, weather forecast, etc.This is the Internet of Things at work.
The idea of devices connecting to each other and sharing data is not new and various sectors of industry have been implementing their own proprietary networks to solve their own problems for many years. Industrial Automation has been reliant on a host of networked PLCs and HMIs for a long time now and those players providing solutions to large manufacturing companies have no doubt done very well from those bespoke installs. What is changing now is that cost, size and power consumption are all reducing for small microprocessors whilst their performance continues to increase – all of which is opening up new and varied uses for essentially, tiny computers.
However you define IoT, everyone is agreed that it will result in a massive explosion in additional devices added to the network which it is predicted, will have to quadruple to nearly a zettabyte at 966 exabytes per year according to Cisco. This assumes a doubling of both people accessing the internet and a doubling of internet connections – most of which are predicted to be M2M connections up from a few hundred million today..The CTO of Ericsson is also famously quoted as saying that by 2020 there will be 50 billion connected devices on the internet which approximates to about 6 connections for every person on earthOh and if you are wondering what a zettabyte is.....
.....then this should helpHowever, for this massive growth in connected devices and sensors, there are a few obstacles that need to be overcome to really enable that to happen.
Once connectivity can be established, then a whole host of inanimate objects could provide useful information to update people as they pass by. This could range from “live” photos, power consumption readings, new emails, weather forecast or the status of your electric car charging in the garage. The information could even be specific to that individual if they can also be sensed by something in the room. Now these are pretty futuristic applications but they stress the point that the technology is now small enough to be embedded into everyday objects and with battery powered connectivity and thin film screens they aren’t that far fetched
As parents, we all face the challenge of allowing our children to go off and play and explore their environment but at the same time wanting to know that they are safe. Tracking can be done by mobile phone but not workable for younger children or pets. Trackable tags could be used to establish location of the wearer and flag if they leave a predefined GPS boundary. The same technology could also be used to protect the elderly and the vulnerable in a passive manner
On the slide we can see an information display built into a bathroom mirror that is also pulling data from other sources.It can recognise the user using facial recognition and display information pertinent to them – such as messages from their doctor and the ability to order a re-prescription of their medication. It can also warn them if the water is too hot or help them save water by monitoring use.Notice the way it recognises when the medicine has been taken by asking the individual to touch the medicine bottle to the glass where its RFID tag is read.This concept could also apply to digital signage where advertising or content could be tailored to the individual based on location, sex or other demographic information
If the patient does have to go into hospital then it is absolutely critical that all treatment, medication and care is specific to that patient.A tablet based platform can identify a patient based on a combination of rfid bracelet, facial recognition and some biometric feature such as a heartbeat. This ensures that patients do not get mixed up and the wrong dosage of medication can be all but eliminated.A full patient record can now be displayed so that appropriate care can be administered irrespective of the personal history of the care giver, nurse or doctor. Wireless sensors on the patient allow real time data to be collected and emergencies to be averted.Again, this scenario seems futuristic but all the technology already exists – it just needs to be implemented by one of you.
Medical devices are becoming connected and the efforts of the Continua alliance are helping to standardise the way devices talk to each other.Here, we are showing how a blood pressure cuff can transmit its data in real time to a data gathering device. In this example, it is a handheld device like a mobile phone but it could as easily be a health monitoring hub in the man’s home that is designed specifically to help him manage a chronic illness at home. Lot’s of innovation is being made in this area and various players such as Microsoft, Freescale and QNX are all investing heavily into these use casesOther devices that are getting the wireless connected treatment include pulse oxymeters, glucometers, weighing scales and thermometers. The data is collected on a daily basis and can then be fed to a patient record. The key here is that if any of the readings are unusual or start trending in a negative direction, the doctor can be alerted before it turns into an acute episode and the patient has to be rushed into hospital.Personal health monitoring allows the patient to sstay at home with little interference from medical staff and the demands on acute care such as hospitals is reduced hence saving money. Consequentially, there is a huge amount of investment going into this space and could be a revenue opporunity for all of you.
The technology is now available to build a battery powered sensor that provides a regular stream of data on light levels, humidity and temperature as an example, as part of, say, a building monitoring systemthe challenges that will hinder massive proliferation are really about solving the connectivity challenges, increasing the native security of these devices and simplifying the development of the sensors and devices. I would like to look at each of these in turn but before I do, let’s look at some of the applications that could be enabled if these challenges are addressed.
So how are all these scenarios going to be enabled if they are not happening today? The key in my mind is connectivity. We see millions of smartmeters being deployed worldwide as part of a push to reduce our energy consumption. To understand how we use energy and where efficiencies can be made, we need to monitor how it is used, where it is used and how to control its use at a granular level. This requires sensors and machines to constantly report what they are doing so that intelligent decisions can be made. So what are our connectivity options?
An obvious choice for connectivity outside the home or office is to use cellular. It is pretty ubiquitous in most countries and makes deploying a device relatively straightforward as it can be pre-configured before being placed in situ with a SIM card and data connection. The current challenges for cellular are of cost, size and blackspots.The cheapest modules for M2M are based on 2G technology such as GPRS or EDGE but the 2G spectrum is not guaranteed out into the future. We are already seeing Japan look to turn off 2G and repurpose the spectrum and other countries could follow. This makes 3G the natural cellular connection but the high cost of modules and data plans make this economically unviable for many IoT applications. What is needed is a low cost module and data tarrif focused on low bandwidth asynchronous applications that do not trigger the significant licensing costs that are often payable to IP holders.One option would be to develop a cellular radio that has been considerably cut down and had the voice, video, MMS, etc. IP blocks removed and therefore would trigger less royalties. A speed restricted data tarrif could be devised to keep costs to a minimum and cellular as a terminal device could become more competitive.However, even if the cost can come right down, the size of modules does not work with the idea of large amounts of low cost (think $10 here) sensors nor does it solve the issue of cellular blackspots, particularly inside industrial and residential buildings.
For buildings and concentrated areas where one entity has control, it is possible to build wireless local area networks using either 802.11 technologies such as Wifi or low power short range networks built on 802.15.4 radios such as Zigbee or 6LoPAN. The main challenge here is around deployment and activation. All the local wireless networks require one device to “pair” in some way with another. This ranges from sharing an encryption key as in the case of WPA protected WLAN or identifying a device on a Zigbee network. There is also a range issue for larger sites and external 3rd parties providing a service into that environment often are not comfortable with relying on someone else’s network to guarantee their solution
So we have a gap....how can we attach small cheap devices to the internet that are easy to deploy, are not restricted to short range and overcome the limitations of cellular as discussed before. Well the current front runner looks to be the use of whitespace. This is the use of the spectrum between the digital TV channels to send small amounts of data over large distances.I am not going to explain the ins and outs of whitespace as it has probably been covered by William Webb from Neul already but suffice to say that it looks like it may be possible to build sub $2 modules that can last 10 years on a button battery but still have a range of kilometres, not meters, and have the attenuation characteristics that gets them through walls into cellars and storage containers.This could be a key technology for the deployment of low cost sensors on a large scale.
As we build out these connected systems, we are exposing multiple additional points of attack and entry into the networked system We need to ensure that the sensor or device is who it says it is, that it hasn’t been cloned or tampered with and that any data it sends and receives is encrypted from prying eyes.ARM has built this kind of security technology into its application processors that are in your mobile phones but the challenge is to enable a similar type of trusted framework for the very small and cheap microcontrollers that will power the sensors and devices we are talking about here.At a high level, we need to ensure that the device can boot securely, that the encryption keys are only accessible by a trusted and hard coded part of the device and that software stored on the device is protected from interference. This all has to be achieved whilst enabling the device to be remotely managed and potentially upgraded at some point in the future.ARM is working on this challenge in conjunction with some of our partners in order to enhance the inherent security of future sensors and devices
Th third challenge i have highlighted is how to ease the burden of software development. It is now well established that software development costs are now by far the biggest cost element of any new product development and making it easier to add sensors to your platform will enhance its capability and increase the deployment of sensors as a whole.One option is to use the mbed developer platform. Please excuse the blatant plug!Low cost reference platforms can be easily integrated with sensors and other interfaces and then programmed using online tools to get a working prototype. Various wireless options are available - wireless LAN such as 802.11bgn or low power connections based on a 802.15.4 radio such as Zigbee or 6LowPAN. cellular modem or some type of whitespace connection will be added in the future.Not just for experiments. mbed based sensor attached to pressure, flow and vibration sensors on Grundfos pump These sensors are feeding a regular stream of data to a cloud based application that can trigger actions to reduce the amount of electricity and water the pump is using dependent on the loading of the system as a whole.Preventative maintenance
And you don’t have to do this on your own – there are many companies out there that can help you build the right solution for your customersThe ARM connected community has over 900 members now and they range from silicon partners to design support companies and also software suppliers, training companies and OEMsIf you are looking to expand your product or service offering then consider registering with ARM’s connected community and you may find someone who can help
Summarise:Talked about how the mobile industry is driving performance per milliwatt innovation that is benefitting the whole embedded computing industryWe have looked at the Internet of Things or Intelligent Systems and how connected devices providing regular or even real time data can enhance existing embedded solutions.And we have looked at some medical focused scenarios that could become additional revenue generating opportunities for you in this room.Thank you very much.