The Internet of Everything

The Internet Today

It’s easy to forget the internet is not a single network. The internet is made up of millions of publicly accessible, interconnected, local networks which which serve over a billion client devices overall. Historically the internet has been viewed as essentially made up of clients and servers, where, for instance, your client laptop or smartphone connects to server to view an organisation’s website, retrieve files or run queries on an app. But this server/client view of the internet is on the cusp of changing in a fundamental way. Today we can now think of the internet in three layers:

• The Core Internet – This has millions of nodes. It is made up of the servers and routers themselves. Client devices use these servers to access data and run server/web applications. The physical composition of the core internet changes relatively slowly and has extremely high capacity. There is an increasing trend for these servers to be based in huge purpose built datacenters. The advent of cloud computing has brought down the barrier of entry to the SME and consumer market to store their own data and run apps in such facilities.
• The Fringe Internet – This now has billions of nodes. It is made up of the client computers (desktops, laptops, tablets and smartphones) used by the end users. These are people centric devices. The amount of devices in the fringe internet is limited to a few billion as the devices themselves require people to use. The topology of the fringe changes rapidly.
• The Internet of Things (IoT) – It is made up of IP-enabled, totally embedded applications within devices that connect to the network. This includes sensors, machines, active positioning tags, radio-frequency identification (RFID) readers and building automation equipment to name but a few. Here there is the potential for trillions of nodes. Imagine every device in your home and workplace, every crucial component in an industrial machine, connected to the internet. This layer of the internet is only just emerging and will completely eclipse the internet as we know it today in terms of scale.

The Economic Benefit towards IoT

There are some big economic drivers towards connecting what were previously viewed as independent devices to the internet, namely: 1) price reduction and standardisation of key hardware and software components and 2) the integration of Big Data/Analytics from which companies can increase revenues and reduce inefficiencies. The Internet of Things has applications in smart metering, industrial automation, logistics, transportation, health, insurance and building automation. Cisco Systems estimates the market value (net profit) in the private sector to be worth $14.4 trillion over the course of the next decade. Putting that in perspective: that’s roughly the GDP of Australia every year for the next 10 years.

Business leaders see operational efficiency, increased customer service and better internal collaboration to be the biggest value drivers:

Industry Examples

General Electric is a good example of a company that is making early headwind into this new area. GE has invested $105 million dollars for a 10% stake in EMC’s “start-up” Pivotal to try and exploit the Big Data/Analytic opportunity that IoT brings. It is also going to market directly with its Predix software suite which is designed to build applications for any system or machine, such as jet engines or deep sea blow-out preventers, and remotely manage them. One of the Predix products, Taleris, which was developed in partnership with Accenture, uses patented algorithms to monitor data collected from aircraft equipment and airline systems to prevent operational disruptions, avoiding delayed departures and flight cancellations. Predix will likely close $1 billion worth of revenue for the company by the end of 2014. At the consumer end of the spectrum GE have launched Link - LED lightbulbs that you can remotely control from anywhere via the mobile app Wink. The app can also control other home devices – locks, blinds, air conditioning etc through the internet.

Technical Foundations

The foundations on which the internet was built haven’t changed much since its inception: In 1974 Vint Cerf and Bob Kahn put forward a model to allow network communication; today it’s common for the whole Internet Protocol Suite to be named after the TCP/ IP protocols they developed. The TCP/IP communication stack allows for a client/server model to work very well.

But what works well for traditional client/servers does not work too well in a world where the internet is completely ubiquitous. A shift away from traditional protocols associated with TCP/IP has been a long time coming for one very simple reason: the world has run out of internet addresses...

Every device connected to a network needs an IP address. These addresses come in a certain format: IPv4, which uses 32-bit (four-byte) addresses, this format limits the address space to just over 4 billion (2^32) addresses. Top-level exhaustion of IPv4 addresses occurred on 31 January 2011 when ICANN gave out the last of the top level numbers to the 5 Regional Internet Registries (RIRs). Each RIR is responsible for giving out IP addresses within a continent (RIRs give addresses to LIRs - companies like BT or Virgin, or other large corporations or institutions such as universities).

While certain protocols like DHCP have helped in making address allocation more efficient they still cannot delay the inevitable: IPv4 addresses will completely run out in the next few years. A new standard system for numbering IP addresses has emerged: IPv6, a hexadecimal system that allows for 128 bit addresses or a total of 2^128 (3.4 × 10^38) addresses. That’s quite a few to go around… an address for every atom that covers the surface of the earth.

Today most operating systems support IPv6, as do most routers and the amount of traffic using IPv6 is growing hugely (1400% annual growth the best estimate around). Surely now that all those extra addresses are available IoT is set to take off for trillions of devices? Well…not quite.

Up until now the devices that have been able participate natively with the internet have been relatively powerful machines with an operating system (laptops, smartphones even “things” like wearables etc.) to deal with the complexity and maintainability of all the protocols that take part in internet protocol stack. But the vast majority of devices that will be connected to the internet as part of IoT will be very different: Most embedded devices have just 8 or 16 bit processors, with very little memory , have sleep patterns to keep low duty cycles, and that use low frequency radio to communicate rather than Ethernet or standard WiFi. RFID tags are a good example of what IoT hopes to encapsulate onto a network.

Having a vast mesh network of these devices trying to communicate over low frequency radio using TCP, HTTP, SOAP or XML, all of which require complex transmission would cripple the IoT network - it isn’t going to work. There are other problems including multicasting and the constantly changing connection topology of IoT networks.

It’s not quite ready for wide market adoption but companies like ARM, Freescale and Amtel are investing heavily in low cost chips that communicate using newly developed protocol standards that integrate fully into more traditional networks. Developments in this area promise to allow for even the smallest devices to participate in the internet. It’s estimated that 99.4% of physical objects that may one day be part of the Internet are still unconnected.

Mobile Networks
The Internet of Things will also transform mobile networks. The next generation of cellular networks, 5G, looks set to be very different from the 3G and 4G networks that have preceded it. While 4G is very well-suited for mobile broadband activities such as web browsing, video streaming, and photo uploading from smartphones, it is not the optimal solution for the Internet of Things, which requires connection to many more nodes, with less data traffic per node but more data traffic overall. Huawei, the largest telecommunications equipment maker in the world, has indicated 5G will have to be far more flexible then 4G and include capacity and connections for at least 100 billion devices. Due to these considerations the wait for 5G will be slightly longer than previous network iterations: the first 5G networks will not be deployed until at least 2020.

Big Data/Analytics
Big Data/Analytics is one of the key enablers of IoT. It allows all information from every device to be collected, stored, analysed and presented succinctly for end users. Intel believe that the Internet of Things will generate far more real-time data than smartphones today, which is more about simply uploading pictures stored in datacenters without much real-time processing.

Big data isn't just lots of data: Big data is typified by the volume, variety, velocity and in some cases veracity of the data (The 3 (or 4) V's).Collected and stored data only gains value once it is analysed and acted upon; the ability to do this in real time is crucial to IoT in particular. Traditional relationship databases like those based on SQL and it's variants are unable to handle Big Data effectively. When Google tried to index all the worlds information they simply could not do it using standard databasing. A relatively common theme among the newer database technologies is that some of the rules for database integrity are relaxed so that the data volume or variety can be managed more easily. Hadoop is probably the most notable example of this.

Today, many enterprise IT architectures are not optimized to either ingest this volume/type of data or make effective use of the resulting information. DHL, which is looking at how best to implement big data in logistics, conducted a survey and found most respondents didn't have a strategy to address Big Data.

Most data in an business is unstructured: On average unstructured data is 3 times as large as structured data, moreover unstructured data is growing 5 times as fast. Data from business partners (suppliers, contractors etc.) such as location based data or financial/credit data is on average 90% unstructured and more widely accessible data through things like YouTube and Google are completely unstructured. In parallel with this the emergence of IoT is also having an effect: 40% of all data is expected to be telemetry based by the end of the decade. Despite all this most businesses only make decisions on their structured data contained in their standard database. If businesses fail to grapple with the problem the result is a tunnel focused view and a failure to react effectively to changing market conditions.

It follows that IT companies that can offer new data store technologies, middleware/intergration and predicative analysis look set to reap the biggest rewards by bringing transformation to businesses.

Long Term Implications

IT can be viewed as having three stages of history: The mainframe era, dominated by the likes of IBM, the server/client era, dominated by Microsoft, and now the cloud era, where companies like Apple, Amazon and Google are vying for dominance. IoT will play a major part in industry disruption, and it's impact on our society could be huge.

By 2040, the world’s population is expected to rise to 9 billion increasing the demand for energy, water, and food. The Internet of Things could provide means to improve efficiency and minimize waste. Smart meters will increase the amount of real-time information provided to system operators and end users alike allowing the possibility for energy and fuel consumption to fall.

There are also positive implications for food production. Data analytics and algorithms based on famers’ historical plot-by-plot yield performance, combined with highly localized historical temperature and precipitation data, should give farmers better information with which to make their decisions. There are three ways in which this could help to improve yields: 1) optimization of hybrid seed choice and plant population; 2) optimization of fertilizer application rates; and 3) minimization of the impact of adverse weather events.

For all these positives there are areas of caution. The Internet of Things is essentially an army of trillions of tiny robots making our lives easier. But these robots will be doing jobs formerly done by people. For example, in retail, cashiers and warehouse staff both could potentially be replaced by connected devices. In the leisure industry,we may ultimately have virtual concierges in hotels and guides in museums. Whilst there is a positive cost impact for organisations through the reduce staff costs, there is also the wider economic threat of higher unemployment rates. Whether sufficient new industries and types of jobs develop to absorb some of the newly idled workforce only time will tell.