Big Data Challenges

Big Data

Big Data has been touted as the answer to many problems. Currently many organisations are collecting everything they can get their hands on in order to try and make sense of it, either now, or eventually. And then of course to leverage that for profit, advantage, protection, whatever…

 

But there is a massive flaw in one aspect of this. For instance, you can determine that there is a statistical correlation between two sets of numbers, but that doesn’t mean it is causative.

Tyler Vigen

Tyler Vigen

Tyler Vigen has written Spurious Correlations to demonstrate just this issue. Below are 2 examples.

Spurious Correlation #1

Spurious Correlation #1

So it looks like the US spending on science correlates with suicides. But does it cause it? Or is this merely a coincidence?

 

Spurious Correlation #2

Spurious Correlation #2

Same here. Would suggest spelling bees stick to short words only to limit the damage they could be doing.

 

And this is what we need to be careful of. Now this doesn’t mean our own thinking can’t be fooled. Rolf Dobelli in The Art of Thinking Clearly documents 99 more common cognitive biases we can suffer from. Think of these as bugs in our thinking algorithms. An example is Confirmation Bias. Most people believe they are above average. Nearly half of them are wrong. This is called Illusory Superiority.

 

So the danger is concluding there is a cause and effect relationship when there is not. Leading to decisions being made and outcomes not working out as expected. There are no simple answers here. The above examples are selected because it is easy to conclude there isn’t a direct relationship. It is harder to discern when the data is big and an auto-correlating algorithm is coming to its own conclusions.

 

The Art of Thinking Clearly

The Art of Thinking Clearly

And to round off, here is Rolf Dobelli talking about The Art of Thinking Clearly.

 

 

Successful Endeavours specialise in Electronics Design and Embedded Software Development, focusing on products that are intended to be Made In AustraliaRay Keefe has developed market leading electronics products in Australia for more than 30 years. This post is Copyright © 2018 Successful Endeavours Pty Ltd.

5G and IoT

5G for IoT

Thanks to the team at VDC Research who compile some very useful information on Embedded and IoT (Internet of Things) trends. It is free to join and the deal is that you contribute to their surveys in order to get access to some reports for free. They also do detailed reports for business purposes which are available for purchase.

 

VDC Research

VDC Research

The following 5G IoT Infographic was put together by them to show the progression of 5G cellular or Mobile Communications in terms of its impact in the Embedded Systems and IoT space. If you click on it you will get a cleaner version to look at and you’ll probably want to zoom in a bit.

 

5G IoT Infographic

5G IoT Infographic

I was interested to see that there are still no fully confirmed standards for 5G. And my previous post on Cellular IoT Communications shows this to be a trend where NB-IoT is still being ratified even though there are chip sets on the market. It is also sobering to think about where all the data will get stored as devices running Gb/sec data streams will have to be sending it somewhere. Big Data keeps getting bigger.

 

Successful Endeavours specialise in Electronics Design and Embedded Software Development, focusing on products that are intended to be Made In Australia. Ray Keefe has developed market leading electronics products in Australia for more than 30 years. This post is Copyright © 2017 Successful Endeavours Pty Ltd.

Cellular IoT Communications

LPWAN = Low Power Wide Area Network

LPWAN is typically thought about as cellular data networks but that involves a contradiction since cellular and low power are inherently in conflict with each other. For instance, a standard 3G or 4G cellular modem will have a peak current draw of up to 2A during transmission and needs to be carefully power managed if running from batteries. This has meant that a 10 year operating life from a primary cell battery either needs a huge primary cell or very infrequent communications. So what are the alternatives?

 

In IoT Versus M2M we looked at how the real benefit of IoT (Internet of Things) is that rather than a single Machine to Machine link being established, there are now multiple devices connected via shared web services and their combined data is being used to create extra value, and particularly if Big Data analytics is added to the mix.

 

SigFox Logo

SigFox Logo

LoRa Alliance

LoRa Alliance

There is also a lot of potential disruption in this. LoRa and SigFox are both looking to provide lower cost networks to replace dependency on cellular network operators for coverage and also address the power consumption problem. There is an excellent comparison of these 2 systems in SigFox versus LoRa. And both are trying to disrupt existing cellular network providers. An overall view at available at NB-IoT versus LoRa versus SigFox.

 

NB-IoT

Which introduces Narrow Band IoT or NB-IoT as it is now commonly abbreviated to. Just to continue the confusion of acronyms, it is also called CAT-NB and CAT-NB1. There is a detailed view of this technology and its likely long term adoption at NB-Iot is dead – Long live NB-IoT.

 

The summary is that NB-IoT is too late to market and requires too much equipment changeover to win the early adopter market, especially in the USA, but will win in the long term. In the interim there is a host of other options also being developed. The cellular network operators have realised, at least 5 years too late, that their business and technology models were both under attack simultaneously. This is a particularly dangerous form of disruption.

 

Hardware is now becoming available and China adoption of NB-IoT makes them the  main early adopter market.

 

Quectel BC95 NB-IoT Module

Quectel BC95 NB-IoT Module

u-blox SARA-N2 NB-IoT Module

u-blox SARA-N2 NB-IoT Module

Low Power Cellular

So if up until now, low power and cellular were not usually compatible concepts, what is changing to address that?

 

To reduce power consumption, you have to have one or more of the following:

 

  • reduce transmit power
  • increase receiver sensitivity
  • reduce transmit duration
  • increase transmit interval
  • reduce network registration time
  • reduce data rate

 

Some of these can be mutually exclusive. However the key elements that are working together is to reduce the data rate and use a modulation scheme that means the transmitter power can be reduced. LoRa does this very well and NB-IoT is looking to achieve a similar thing. There are trade-offs and the lower data rate for NB-IoT means it is best suited to very small packets. CAT-M1 will require less power for larger packets because the faster data rate means the transmit time is a lot shorter.

 

Low Cost Cellular

So we have looked at the power consumption angle. How about cost and business model. And there are two aspects to cost. There is the hardware cost and there is a the network operations cost. To reduce cost you have to do one or more of the following:

 

  • reduce silicon and software protocol stack complexity
  • high volume production allows economies of scale for hardware
  • increase the number of channels available in the network
  • increase the number of simultaneous connections in the network
  • reduce margins

 

Both SigFox and NB-IoT aim to make the end device hardware cost as low as possible. In the case of NB-IoT and CAT-M1 the channel bandwidth can be reduced and so the same bandwidth can support multiple devices instead of just one. The power level in the device transmitter is reduced by reducing the bandwidth and data rate. As an example, a CAT-M1 module has a peak transmitter current draw of 500mA which is a factor of 4 lower than CAT-1. So low cost and low power can go together very well.

 

The graph below shows how the various cellular standards relate to each other.

 

Cellular IoT standards and how they relate

Cellular IoT standards and how they relate

IoT Deployment Options

We have been using standard 3G/4G Cellular modems for our broadly distributed IoT offerings. As of the end of this month, we ship our first CAT-1 based offerings. These have the advantage of supporting both 4G with fall back to 3G. Although NB-IoT hardware is available now from both Quectel and u-blox, the networks in Australia don’t yet support it. And while NB-IoT is ideal for fixed location assets, we also do mobile systems so these need to be CAT-M1 once it is available.

 

CAT-M1 is expected to be available in Australia on the Telstra network around September 2017. I am also taking this as meaning that NB-IoT is 2018 or possibly even longer. So we plan to move to CAT-M1 as soon as it is available. The modules are expected to be available about the same time as the network upgrades.

 

Here are some CAT-1 and CAT-M1 offerings from Quectel and u-blox.

 

Quectel BG96 CAT-M1 Module

Quectel BG96 CAT-M1 Module

Quectel EC21 CAT-1 Module

Quectel EC21 CAT-1 Module

 

The Quectel EC21 is what we are deploying in our units later this month.

 

u-blox LARA-R2 CAT-1 Module

u-blox LARA-R2 CAT-1 Module

 

u-blox SARA-R404M CAT-M1 Module

u-blox SARA-R404M CAT-M1 Module

IoT Network Upgrades

Ericsson have announced the roll out plans for the Telstra Network CAT-M1 capability.

 

And Telstra have announced their own Telstra IoT Network Plans.

 

That is the overall Telstra road map. Summary:

CAT-1 now
CAT-M1 by September
NB-IoT sometime after that but no dates yet

 

Other carriers will follow although Vodafone are well placed to introduce NB-IoT first as they have Software Defined Radio base stations from Huawei and so can roll it out as a software update.

 

Successful Endeavours specialise in Electronics Design and Embedded Software Development, focusing on products that are intended to be Made In Australia. Ray Keefe has developed market leading electronics products in Australia for more than 30 years. This post is Copyright © 2017 Successful Endeavours Pty Ltd.

IoT versus M2M

IoT versus M2M

M2M or Machine to Machine communications have been around for a long time. So how is the IoT or the Internet of Things different?

 

The picture below shows that basic paradigm shift between the two concepts.

 

IoT versus M2M

IoT versus M2M

The easiest way to think about this is that Machine to Machine communications is a subset of the Internet of Things. IoT can make use of M2M but M2M on its own is not sufficient to create IoT.

 

On the left you see that you can have specific devices connected to each other via some form of communications. So I can send a Fax or check my fitness level. Even get back data from the Pioneer and Voyage spacecraft as they explore the solar system and eventually head beyond it.

 

One the right, we add ubiquitous communications, cloud services and Big Data correlation and we have a much more power ecosystem that also creates a lot more value. And that is what is driving IoT Growth, the extra value it creates.

 

Successful Endeavours specialise in Electronics Design and Embedded Software Development. Ray Keefe has developed market leading electronics products in Australia for nearly 30 years. This post is Copyright © 2015 Successful Endeavours Pty Ltd.

Technology Creates The Future

How to change the world

To make a significant difference in our world, you either have to do something outstanding or have a lot of influence. Or both. And it will need many people to contribute over the course of the project.

 

So I keep an eye on some of the forums that help people to collaborate in order to facilitate this:

 

 

And I spotted this piece of news that shows a radically different way to deliver remote Internet services.

 

 

The project is a classic example of using Big Data to solve a big problem. The project is of course Project Loon and is intended to connect remote or difficult sites to the Internet including disaster recovery scenarios. The challenge is how to do it without wiring or expensive satellite links and the name suggests it is one of those ideas out of left field. The solution being trialled is a series of balloons floating in the upper atmosphere which provide an intermediate link between ground based transceivers. It uses standard ISM frequencies at 2.4GHz and 5.8GHz and so can operate with standard equipment and delivers GSM levels of Internet speed.

 

The truly brilliant part is taking the winds direction data for the upper stratosphere and using that to control the altitude of the balloons so they stay where they need to be. A fantastic example of gathering and using data on a scale that most of us can barely appreciate.

 

Project Loon Balloon

Project Loon Balloon

 

So well done to Google for taking on such an ambitious project.

 

And they are starting with New Zealand Hinterland Internet.

 

Successful Endeavours specialise in Electronics Design and Embedded Software Development. Ray Keefe has developed market leading electronics products in Australia for nearly 30 years. This post is Copyright © 2013 Successful Endeavours Pty Ltd.