Networking for IoT: The Role of LP WAN, SD WAN and Other New Components


By Trevor Clarke, Tech Research Asia co-founder and director

The new world of the Internet of Things (IoT) is effectively the realisation of a whole gamut of ideas. And more will come. In my view that is a good thing - the ability for anyone to innovate with the current low barrier to entry with technology is exciting. IoT has, however, created an important change in the way organisations need to consider their networking strategy. It’s already a world with multiple new applications and ways of doing things that fundamentally rely on ever- increasing connectivity requirements. Many, if not most, of these new applications have different bandwidth or traffic profiles than their predecessors. There are those with lots of smaller data packets but more frequent traffic (up and down). Some have heavier bandwidth requirements, and some have real-time needs in both urban centres and remote regions. In all of this, there is no one standard approach being taken as the range and type of applications is widely diverse. And it’s all expanding quickly in all manner of ways: data volume, velocity and variety; the number of connected devices and their form factors; and the ways of connecting things.

At the same time as IoT is exploding onto investment agendas, we’ve had the emergence of software defined networking (SDN), and with it SD WAN (or software defined wide area network). This latter approach is raising the question of whether organisations can use consumer-grade internet broadband connections for some, or even all their requirements in a smarter fashion via programmability, instead of more expensive carrier’s MPLS WAN (or Multi-Protocol Label Switching Wide Area Network), which is a distinct private network from the public internet. The chart above highlights the growth in fixed-broadband connectivity across Asia Pacific. This will only improve as time goes by and as quality of service improves, so too will the potential for organisations to use SD WAN.

Further, in addition to the growth and steady adoption of centralised cloud computing services and marketplaces (that frequently enable IoT projects) that have been established in almost all countries world-wide over the past five years, we are now seeing a return of distributed computing in the contemporary form of edge computing. In effect, the typical hybrid data centre and cloud computing environment – which is arguably the most common or desired set up – are starting to evolve to encompass edge environments as well.

Then throw in the emergence of many low power wide area networking (LP WAN) technologies and protocols being pursued and jockeying for pole position – along with a host of other developments such as the possible future of 5G and satellite broadband – and you have a decidedly animatedly evolving state of affairs. A situation that offers decision makers far greater choice, but also potential complexity in what is effectively an unfamiliar environment.

We have not been here before. Let’s just repeat that – we have not been here before. This is, for all intents and purposes, a new frontier that requires new thinking and strategy.

Combined it is TRA’s view that this is a noteworthy shift in what is required to ensure networking performance in organisations of all types. It suggests a need to re-evaluate the networking status quo to ask whether new approaches would serve us better as we head deeper into the world of IoT.

So why should we now consider change?

This article is intended as a general overview for business and IT leaders who we believe should be looking into things like LP WAN and SD WAN as they consider or move into IoT-type projects. (It's not a technical guide but does have some technical information!) From the customer and end devices through to clouds and datacentres we have never had more options. The following are some of the important trends influencing networking developments today. They are presented in no particular order and in summary only:

Customer expectations: Being always on, always connected and having the flexibility to control an experience or interaction with an organisation is the base expectation of customers today. This has only emerged in the past five years, but will intensify.

Cloud computing: Using multiple different cloud computing services that may be located in multiple places around the world, in addition to applications located on-premises.

Expansion of Network Infrastructure: Investment in the addition of new bandwidth and backhaul from the customer premises to undersea fibre cables is ongoing. This also includes experimental approaches and contemporary satellite connectivity.

Expansion in types of connectivity: Particularly with IoT projects there has and continues to be an increase in the number of networking protocols and standards. In short, the way we could connect is increasingly diversifying.

Expansion in what is being connected: It’s why we call it the internet of “things”. You take any physical item you can think of from the clothes you wear to the machines you use and there is likely to be someone around the world trying to put a sensor in it and connect it to a network.

Data’s Role: We know the volume, variety and velocity of data is increasing. Importantly the value of data is also increasingly being recognised and leveraged.

Bandwidth Costs: Yes, it does. And that cost is rarely coming down for the average organisation in Asia Pacific. On the contrary, it is often increasing.

IoT Diversity: Growth in the variety of IoT applications is at the start of its expected uptick in markets around the world.

Software Defined X: Abstracting functionality away from hardware layers to enable things to be programmed.

Enabling Edge: A significant push by suppliers (often for sound technical and business reasons) to educate and stimulate the market around edge computing.

Blockchain: The technology underpinning Bitcoin that could also be the basis for a decentralisation movement in networking.

Makers Movement: A revolution in hardware spurred by the low cost of components, 3D printing, software platforms, and improvement in skills. This includes networking hardware.

Spectrum and standards: The reality that radio spectrum is a managed and finite resource. Once which those who have paid for spectrum licenses will work as hard as they can to achieve a strong return on investment.

Security: It will not get any easier to protect an organisation from threats nor respond to incidents once they happen. Network-based attacks, like distributed denial of service (DDoS) attacks, are now rentable and increasingly potent.

In TRA’s view, the above should, either independently or in aggregate, convince your decision makers to re-evaluate your networking environment or at the least give it a health check. This is particularly the case if your organisation has plans for pursuing IoT projects.

In addition to the market dynamics and drivers we briefly introduced, many IoT projects differ widely in their architecture and required skills from their predecessors in IT and the common networking approaches introduced above. This is as true in a white-collar office environment (aka smart buildings) as it is on a farm. IoT projects may, for example, be attempting to digitally connect objects which were inherently analogue and which we’d never thought of connecting before (i.e. a fence or gate). They may be requiring network connectivity in places were before there was no need or over a far greater area than the past. Moreover, the required performance levels of the application and network can be either far lower, or far higher depending on the desired outcomes and nature of the application.

There is no one single way that IoT projects are different from what most of the familiar applications organisations around the world have deployed and managed to date. Rather, there are likely to be many differences and choosing the appropriate network connectivity should be considered on a case-by-case basis with as few pre-conceived notions influencing decisions as possible. Because on the networking side of this are a range of new or evolved components or options to consider, with many of them developing and maturing quickly. We introduce some, but far from all (for instance we deliberately leave out cellular and fixed-line broadband), of them below. In TRA’s view the following are some of the more significant areas in IoT networking today.

Low Power Wide Area Networks (LP WAN)

LP WAN is the current poster child of IoT project networking. For good reason. The emergence of more and more “things” that are laden with sensors and connected in more and more diverse places in order to send the sensor data somewhere for analysis and action has required new approaches. The sheer scale of many projects meant we simply couldn’t continue to rely on traditional WAN set ups and the related market services. Most networks, such as terrestrial, cellular, and satellite broadband were not built for use cases that involved tens or hundreds to thousands of IoT sensors or devices requiring connectivity. Thus, there are now several different LP WAN protocols, each with their own burgeoning industry of hardware makers and service providers jockeying to be the “one” preferred LP WAN platform.

As the name suggests, some of the features of the LP WAN family are that:

  • They support long-range wireless transmission of data typically small packet sizes but can be theoretically up to 40 to 50km

  • Provide excellent wide area geographic coverage in both remote and urban areas

  • Are suitable when short range options like ZigBee or Bluetooth, for example, can’t be used, and when cost prohibits using cellular or satellite services

  • Have low bandwidth but can be symmetric data flows (up/down).

  • Use low cost devices that use low power and can be left for long periods of time; `10 years in some circumstances.

  • Mostly uses unlicensed spectrum and thus can suffer from the congestion and noise issues that this entails. Yet they are also, therefore, usually significantly lower cost overall

  • Do not have IP capabilities and thus require a gateway to connect to the internet or MPLS networks, or act as a closed network.

  • Can have thousands of devices connected to one gateway

  • Are limited in how many times per day they can send data and suffer from high packet loss

  • Mostly use a star of star topologies, but can also use mesh

  • Limited when it comes to applications that require accurate localisation capabilities

  • Are the subject of lots of marketing spin from LP WAN providers and their competitors (such as telco providers that see it as a potential risk)

Some example use cases for LP WAN include:

  • parking garage sensors

  • water or air quality meters

  • smart meters for utilities

  • monitoring of infrastructure or machinery

  • soil monitoring

  • tracking animals

  • monitoring bins or other stationary objects

  • logistics and asset tracking

  • predictive maintenance

  • IFTTT-type scenarios (“If This Then That”)

Each of the different LP WAN protocols have their pros and cons. The main types available include:

Short Range Wireless

Prior to the recent interest in LP WAN and the wide area coverage this offers, the buzz was all about other mesh networks and ways of connecting that give short range smarts. In particular, along with ongoing and long-standing interest in WiFi and RFID, considerable attention and investment was allocated to:

  • Bluetooth Low Energy: An industry standard found in many consumer devices today. Became popular with the use of beacon technology in retail and other consumer-facing situations where location based services were desired and smartphones were used.

  • ZigBee and Z-wave: Two popular and competing short range wireless protocols that have smart home device ecosystems in place. ZigBee has also been used in several industrial IoT projects. For example, it has been combined with wearable devices to monitor worker occupational health and safety on construction and mine sites.

  • NFC: Near Field Communications, as the name implies, is a wireless protocol frequently used between devices in close proximity, such as a smartphone and a payment terminal.

Satellite

Satellite is a staple in remote locations around the world where cellular connectivity is absent. It is also used for several in-motion connectivity IoT (and M2M) use cases such as logistics and marine or shipping containers both within and across borders. Not only are satellite services used for navigation (such as in CORS networking) but are also used for data transmission in IoT projects. There are many national and international providers but pricing and latency remain issues, especially for IoT projects that may have hundreds or thousands of devices being connected.

Recent developments, are however, pointing to an exciting future. Firstly, the manufacturing of satellites is improving: they are far smaller, cheaper, and faster to produce. Second is the investment being made in putting these Nano-satellites into low earth orbit to provide affordable and comparatively good broadband connectivity. SpaceX is one example of a firm pursuing this goal, and OneWeb is another. Combined with IoT developments, TRA expects many of these newer satellite services to change what is possible in remote locations.

SD WAN

Another important change in the market has been the emergence of software defined networking (SDN) and with it SD WAN. SD-WAN is being positioned as the next technological step in the provision of WAN services. Networking functionality has not changed much in regards to architecture since its inception in the 1980s. The approach to networking has been customised configuration of each device (routers and switches), connected with fixed links, as this provided predictable pricing over a medium that was considered significantly more stable and secure than the Internet. Legacy applications as well demanded stable networks to function successfully, due to the intolerance of networking software within operating systems, and the application itself.

With SD-WAN, the approach takes advantage of the Internet connectivity, bandwidth availability, and stability improvements, as well ensures applications which can be sensitive to network fluctuations, have a reliable transmission path for data. The proposition of SD-WAN is that it is meant to allow corporations to maximise networking budgets, increase bandwidth, and reduce latency across networks, without impacting the functionality of applications.

Put simply, SD-WAN places software-based intelligence into the network routers so that it can identify different types of application data and use ‘performance-based routing’ to select the best method of sending the data to its destination. MPLS does not route traffic based on application, but only prioritises application data across a single MPLS link. By using SD-WAN ‘performance-based routing’, traffic is intelligently sent over the Internet or a MPLS link using the most effective network available. So, if an IoT project has large sets of data but does not need the performance of MPLS, it might be more prudent to send it across the Internet using SD WAN.

Others

As we move further and further into the world of IoT, and as efforts continue to connect the unconnected to the internet, innovative network approaches continue to be pursued. Indeed, there is a lot happening. TRA does not expect all of the ideas to come to fruition – many will simply fail and be remembered as novelty projects – but in aggregate, it tells us that different approaches are possible. We don’t need to rely on the ways we have always done things when it comes to IoT. Below are a few approaches that are worth considering in terms of how they might be applied or provide inspiration for solving connectivity issues.

  • Simultaneous Localization and Mapping (SLAM): SLAM is not really a new development, but in the past few years its development has accelerated considerably as a result of its important role in Mars rovers along with the surge of investment in autonomous vehicles. In short, it is the way robots and autonomous vehicles navigate. Using radar, GPS, a bunch of sensors and on-board processing, vehicles or robots (rovers) are able to autonomously navigate their way through unknown terrain by creating maps on the fly. This has significant implications for remote regions where connectivity is limited.

  • Project Loon: Google’s balloons that are being trialled to provide internet access.

  • Aquila: Facebook’s solar powered plane that is hopes will provide broadband to the masses in future.

  • Blockchain-based IoT devices: There are signs of a move to use LP WAN and mesh networking approaches combined with blockchain technologies in the world of IoT. Filament is one start up example that provides a self-aware mesh network with smart contracts (based on blockchain technology).

  • Edge Computing: instead of sending all of the data from IoT projects back to a central data hub (often in a cloud-based IoT store), many organisations are exploring edge computing and locating infrastructure as close to the customer or action as possible. For example, as a store and forward set up co-located with internet gateways for LP WAN-based projects. 

IoT Connectivity Comparisons:

  • The chart below provides a visual means in which to understand the different IoT networking options available to all organisations. On the Y axis is the possible bandwidth and on the X axis is the possible range of each protocol included. It's just a guide. You should always do your due diligence before making any investments. TRA Networking ComparisonFor more information on Tech Research Asia's advisory services and innovation workshops for Asia Pacific enterprise executives please visit www.techresearch.asia.