Following on from our CEO’s recent piece on the Internet of Things, he continues his series of blogs here. This time, Mark Harrop discusses the IoT from a more technical perspective, exploring TCP/IP standards, PPP, and Mesh Networks. 

I recently wrote an article on the subject of the Internet of Things (IoT) titled: The Internet of Things – Transforming Fashion. This was a precursor to WhichPLM’s upcoming 6^th Edition print publication, which is dedicated to analysing the impact that the IoT is likely to have on the fashion and retail industries, and which will include exclusive interviews with the inventor behind the term, as well as other leading industry figures.

After that article was published, I received an avalanche of support, comments, and questions that helped to reinforce something that I and the rest of the WhichPLM team firmly believe: the IoT is set to change our industry, and indeed our entire world, more than any other technological revolution.  So needless to say, I wasn’t short of encouragement and suggestions for my next article in the series!

In my last piece, I discussed what WhichPLM’s analysts consider to be the logical next steps for the IoT in fashion, driven by the significant uptake in fashion-specific PLM solutions (a figure that stands at around 1,900 worldwide according to our latest market analyses) and the slow but steady expansion of their footprint into the broader enterprise.  Regular readers will be familiar with the E-PLM (extended PLM) concept we pioneered some years ago, explaining that, as a backbone for intelligence, innovation and integration, PLM now plays a pivotal role in allowing brands, retailers, and manufacturers to bring on board new processes and supporting technologies – things like enterprise 3D, digital transformation and the IoT operating near and at the edge of the Internet.

Building on this concept, I want to dedicate this second article on the subject of the IoT in fashion to understanding precisely how – beyond the software level, the IoT is likely to actually work within our industry.

The critical, but often overlooked, part of the Internet of Things is the word “Internet”.  People tend to look at it as a closed book, or a kind of magic, forgetting that a tremendously complicated communication network is providing the infrastructure that allows the more obvious and exciting applications to happen.

I’m certainly not claiming to be a network expert, but I did recently set out to better understand the underlying hardware, protocols, and interfaces that make the IoT possible.  And I want to set out my extremely high-level understanding of these in the hope that it will remind fashion organisations of the sheer complexity of the IoT, and spur them into scoping, planning, designing and developing the applications and use cases that will bring them value.

A word I’m going to use a lot is “topology”.  In the IoT sense, this means the way in which parts of a communication network are arranged; it’s a schematic description of the interrelationships between different nodes, components and connecting lines.  In the case of the Internet, there are two ways the word can be used: to describe the physical distribution of cables, relays, cabinets, DNS servers and so on; or to talk about logical (or signal) geometry, the way that the information sent is structured and arranged.

For our purposes, we need to understand some basics of how the Internet is built, using the architecture standard of Transmission Control Protocol – Internet Protocol, or TCP/IP.

Developed by the Department of Defence’s Project Research Agency (ARPA, or later DARPA) TCP/IP is an open network architecture (all solution vendors are free to design and develop solutions that use TCP/IP standards) that’s been adopted as a worldwide network standard and widely deployed on LANs (local area networks) and WANs (wide area networks).

TCP/IP was originally created to connect military networks together; later it was used by government agencies, and finally universities before the Internet was properly commercialised.  As a transmission prototcol, it is robust to failures and flexible enough to be used across hugely diverse networks.  It is, broadly speaking, the best way we have of sending data from one location to another.

But how does information actually travel across cyberspace from one computer to another? The crucial distinction is that it doesn’t make the journey whole.  Information is disassembled into small blocks of data, sent from one computer/device independently to the destination address (computer/device) and then reassembled at the receiving end.

Each of those blocks of data is known as an IP (Internet Protocol) packet.

But while we obviously know, geographically speaking, where each device is, how do the devices themselves identify their place in the network, and do they maintain it indefinitely?  And are there enough places to go around?

Each time you connect to the Internet, your ISP (Internet Service Provider) assigns your computer a unique numerical address. This unique address identifies your computer on the network so that you can request and receive information.  The address is known as an Internet Protocol (IP) address.

When you initiate a request, such as clicking on a link in your Web browser to whichplm.com, the request travels across the Internet in the form of data packets and is stamped with your IP address – i.e. the information knows where it came from, and the response knows where to find you.

And size really does matters when it comes to IP packets: the smaller the better.  Smaller packets are more efficient than large packets; for example a single large packet holding large amounts of data (e.g. lots of text and native images) would take far longer to transmit and with high traffic volumes could quickly bring your network to it’s knees, which is often the case in many fashion businesses that use lots of native file imagery (uncompressed images).  This is what’s known as latency – the inherent speed of your network.  Sending smaller packets of text and compressed images only (up to about 1400 bytes) will have the advantage of moving more traffic at a much faster pace with less chance of creating bottlenecks for the IT department.

But these packets don’t make the journey alone: each IP packet travels the Internet guided by routers that forward the package onto the destination address along the fastest available route. Along the route the packages are audited to ensure that they are complete and then reassembled ready for the receiver to act upon.  To follow our example, a successful request from an HTML browser is processed by a Web server that responds by sending the requested whichplm.com home page, which itself is broken into many small IP packets for the return trip to your computer screen. These rushing data packets make up what people used to call “The Information Superhighway.”

As I said, this is a simplified and high-level overview of how data travels through the Internet, but now I want to talk about how connecting a huge volume – millions upon millions – of new sensors and devices to that network (RFID chips, machine to machine, machine 2 humans communication and so on) are according to the experts necessitating a redesign of both TCP/IP itself, and the topology of our network infrastructures.

The first thing to consider is that not all networks protocols are arranged alike; they can have dramatically different topologies that affect their capabilities, and their suitability to support the IoT.

PPP (Point-to-Point Protocol) is a protocol for communication between two computers/devices using a serial interface, typically a personal computer connected by phone line to a server. For example, your Internet server provider may provide you with a PPP connection so that the provider’s server can respond to your requests, pass them on to the Internet, and forward your requested Internet responses back to you.

The primary limitations of using a one-to-one P2P relationship that exists between two devices; is that the network cannot scale beyond these two nodes. Each point-to-point link is technically called a hop. The hop count is the number of network devices between the starting node (computer/device) and the destination node (computer/device).  The range of the network is therefore limited to one hop, and defined by the transmission range of a single device.

Star Networks are one of the most common computer network topologies. In a star network each computer/device on the network has its own cable or wireless data interchange that connects to a central hub. The hub gateway will send and receive data packets to and from each of the connected devices, including sensors. An example of a Star Network is the WiFi network hub in your house, which is connected to your many devices/computers.

A Mesh Network is a network topology in which each node (computer/device/ sensor) relays data for the network. All mesh nodes cooperate collectively in the distribution of data in the network.

Mesh networks can relay messages using either a flooding technique or a routing technique. With routing, the message is propagated along a path by hopping from node to node until it reaches its final destination. Its main advantage is that the network traffic can be redirected to other nodes if one of the nodes goes down for any reason, using algorithms to define the shortest path to the target. Self-healing allows a routing-based network to operate when a node breaks down or when a connection becomes unreliable with interference or power outages. As a result, the network is typically quite reliable, as there is often more than one path between a source and a destination in the network. Although mostly used in wireless situations, this concept can also apply to wired networks and to software interaction.

Sensor/router nodes, used within the mesh network are smart sensor nodes with repeater/routing capability, in other words smart sensor/chips/routers that will take small data packets called “chirps” (something we will come to shortly) from very simple end devices that will typically send information one-way, often referred to as an arrow.

As you can hopefully see, these examples (taken from a book I recently read on the subject) demonstrate that vendors, brands, retailers and manufacturers alike need to be mindful of not just the standard methods of connectivity, but their cost and practicality in the new emerging world of the IoT.

In order to support the predicted explosion of connected sensors and devices, the network that supports the IoT will need to be both efficient and scalable without becoming prohibitively expensive.

Unlike our current world, governed primarily by smart devices talking to one another, the future of the Internet of Things will be completely different – characterised by very low cost sensors with little memory or processing power (in the case of the magnetically-driven RFID, no power at all) constantly projecting data that they themselves do not understand.

To put this into context, some people tend to think of the IoT as being about smartphones or fitness trackers, when in fact the revolution will be driven by embedded sensors and chips that cost as little as cents to produce, and that are as “dumb” as can be.  These will be extremely simple nodes sending out traditional TCP/IP packets and minute one-way bursts known as “chirps” that are a different prospect entirely.

In a traditional network topology TCP/IP sends and receives relatively heavy data packets with the aim of maintaining reliability, security, encryption, and audits – all geared around ensuring that a packet is received in full.  Much of this heavyweight data transmission – particularly the return leg – becomes redundant when we consider that the largest proportion of nodes in a near-future Internet of Things will be extremely simple, transmitting small amounts of data (shade batching information, footfall counters, SKU details, basic material information, by way of examples) that start at the edges of the network and work their way inwards, like leaves on a tree, following it’s branches and arriving at the centre of the trunk.

IoT “chirps,” on the other hand, are lightweight packets, containing no security, checks, audits or real meaning in and of themselves.  Interpretation only begins at the “interested” Integrator functions that can route the chirp data downwards and upwards.

The IoT, then, unlike the traditional Internet is receiver-centric, not sender-centric.  While a smart device is interested in sending and receiving consistent, uncompromised information, a dumb node (or even a huge network of dumb nodes) simply sends out its status without necessarily getting anything in return. To give this some weight, the number of bytes in a TCP/IP header is around 40 compared to typically 3 bytes in chirp header, making the latter the logical cost effective approach to sharing the data generated by sensors at the edge of the IoT.

And because these chunks of data are so small, the costs of this over-provisioning at the very edge of the IoT are infinitesimal. But the benefits of this sort of scheme are huge. Since no individual message is critical, there’s no need for any error-recovery or integrity-checking overhead (except for the most basic checksum to avoid a garbled message). Each message simply has an address, a short data field, and a checksum. In some ways, these messages are what IP Datagrams were meant to be and more importantly will for the very “first” time will allow retailers, brands, manufactures and the extended supply-chain partners to analyse data to gain important trends and insights.

According to a recent book “Rethinking The Internet of Things” that I read on IoT, data transmission can therefore be lossy (losing data being an acceptable characteristic) and intermittent, so the end devices will be designed to function perfectly well even if they miss sending or receiving data during transmission failures  – the fact is that the data “chirp” feeds will send basic information over and over that will then be aggregated and analysed by algorithms (with human tweaking) to support better decision making.

Topographically speaking, propagator nodes will use their knowledge and filtering of adjacencies to form a near-range picture of the network, locating end devices and nearby propagator nodes. The propagator nodes will intelligently package, filter and prune the various data messages before broadcasting them to adjacent nodes. Using the simple checksum and the “arrow” (single direction) of transmission (toward end devices or toward integrator functions), redundant messages will be trashed.

Propagator nodes will be biased to forward certain predefined valued information in particular directions based on routing instructions passed down from the integrator functions interested in communicating with a particular functional or geographic neighbourhood of end devices. It is the integrator functions that will dictate the overall communications flow based on their needs to get data or set parameters in a neighbourhood of IoT end devices.

While I’ve referenced RFID multiple times, this really is only a small piece of the picture when we consider the huge number of sensors, hardware, software and extended-PLM solutions that could be used to drive value in the design and manufacturing of fashion related products.  So, if we are to build solutions and methods of management to make the IoT a reality then the quicker we start connecting the millions of new nodes, software solutions, machines to machines, machines to humans and digitally transforming our industry the faster we will start to drive the value of joining the dots. My research has also shown me that it will be necessary to define and agree a new low-cost method of sharing packet data well beyond the traditional TCP/IP standards, so lot’s to do for the sensor designers and networking experts racing to deliver their solutions to support the IoT.

Not only does the IoT impact the retail and supply-chain side of fashion business as described in my last article on suggested “use cases”, but let’s not forget the use of IoT related technology to support wearables and “smart” apparel garments which are growing by the day.   It’s not only in the retail stores or factories, but also sensors built into clothing, footwear, watches and accessories, that will connect with the Internet and equipment of all types that will share small bytes of data, resulting in lot’s of new and exciting opportunities!

Regardless of the apparel product or equipment type, smart technology will be collecting, storing and sharing data and will connect to the network as part of the IoT that will create Big Datathat we will need to analyse to make sense of the data and to improve our businesses going forward.

For now, though, keep chirping!

*The images within this article have been taken from WhichPLM’s educational presentations on the Internet of Things.