The Future of Everything: Making Sense of the Sensor Revolution from a Telecoms Perspective
We are in the midst of a revolution in computing, driven by ubiquitous and cheap sensors tied to machine intelligence. In the 1990s the hypertext revolution gained a culturally standardized name, the Web. This 'hypersense' revolution has yet to gather its moniker. One common and sensible framing is the 'Internet of Everything'. It potentially touches nearly every industry and human activity.
At the heart of the Internet of Everything (IoE) is an infrastructure for relaying sensed data and relating it to its meaning. This infrastructure conveys information that is critical to economic success and even safety of life. How should the telecoms industry adapt so it can be ready to supply a platform for a new communications paradigm?
'Wellth' is the new wealth
As a society we are collectively moving beyond the initial Information Technology revolution. This was very much focused on the automation of preexisting business processes. Citizens hold the roles of both consumers and workers. As such they increasingly seek more than mere workplace 'productivity'.
No matter how much material wealth is on offer as the benefit of industrialization, it is of no value if it costs you your health and happiness to produce it. Being endlessly commoditized as labour neglects our human needs for belonging, meaning and connection.
The shift from material accumulation to wellbeing is part of a wider transformation of society via the cheap commodity of ubiquitous information. This parallels the transformations wrought by coal, electricity and petrochemicals. Over time, the economy and social order adjusts to the presumed existence of the new abundant resource.
Embedding information into everything
These industrial revolutions trigger second-order effects. In the case of the automobile, driving rapidly became affordable and vehicles grew ever more dependable. This led to commuting, suburbs, and strip malls. The information revolution and self-driving cars will also have second-order effects that we have yet to experience.
For instance, this cheap information will be used to create mass-customized transport-on-demand services. Your 'autotaxi' may even pre-position itself in anticipation of you summoning it. To do this the service may need to sense whether you are about to go to bed or leave the house.
By their very nature, services that are tailored to individual needs are demand-led and fit-for-purpose. These are implemented by higher-order information systems that model and parameterize services. For example, your airline may look at your travel history and pitch you an annual 'flying plan', with your own unique loyalty program tied to the travel partners you most prefer.
This phenomenon is rather like how flexible manufacturing systems transformed custom engineering using computer numerical controlled (CNC) lathes in the 1980s and 1990s. For it to work, the service delivery engine needs access to personal data to configure the service, so it is just right for you in your present context. Dependable delivery of this capability requires reliable information infrastructure subcomponents.
Engineering for predictable performance
This demand for reliability in turn means that the communications industry needs to engage in an engineering revolution. We must create predictable application outcomes and costs in return for a given level of distributed computing input resources.
Predictability of performance outcomes needs a supporting engineering paradigm and underlying science. This is weak or missing from the current mainstream, which is a huge issue and opportunity. The performance engineering of complex distributed systems is immature compared to established activities like construction and aerospace. We still work to 'unsafety margins', as we can’t accurately model the 'performance cliffs' in advance of deployment.
In solving these difficult engineering problems we will discover that the concepts of 'telecoms' and 'cloud' are mirages. Our historical industry categories obscure a bigger reality: we are all just part of a distributed computing industry. What was once thought of as 'convergence' of telecoms with IT is more of a tumultuous 'collision' of giant ecosystems.
The evolving nature of demand
As the computing and communications infrastructure matures, we will find that there are real IoE revenue opportunities. These will address universal human needs, and tackle 'jobs to be done' previously out of the reach of machines.
To illustrate, we all seek wellbeing and flourishing, yet the current 'sickcare' model of medicine serves us poorly. It treats the human separately from their context, and medical monitoring data is generated at point intervals in institutional settings only once the person is already ill.
Wearable technology will turn the model upside down as we take more personal and proactive responsibility for our own wellbeing. Data will be gathered continuously, and we will own it ourselves. The collective use of that 'hypersense' data will allow us to engineer what has been termed 'hyper wellbeing'.
This example illustrates a paradigm shift in the future use of technology. Productivity is not enough; indeed, from a human wellness perspective, more productivity can be counter-productive. What matters is engineering positive feeling states and beneficial ethical outcomes for individuals, communities and the biosphere.
A 'hypersense' revolution as big as the Web
This paradigm change in purpose is allied to revolutions in transport, energy, education, and more. Each industry is undergoing its own information transformation. These are fuelled by a rapid growth in sensors and sense-making via machine learning. This 'hypersense' revolution is big enough that you can make a good case that 'cognitive' is the new 'mobile'.
To gain a perspective, you can think of the information revolution as being around the same level of development as steam travel by railroad in the mid-19th century. The existence of the information equivalent of high-speed maglev trains travelling at hundreds of miles per hour is unthinkable to most people.
The ability to make sense of the 'hypersense' world enables new forms of contextual computing and communications. The machines can increasingly initiate action in the world on behalf of people. This is a collective phenomenon akin to the arrival of the Web in the 1990s. We might call it the 'Decision Matrix'.
This will put enormous new demands on our networks. It's not just about media like video and virtual reality. The real constraint is our ability to signal and coordinate in near real-time. The consequences of failure are going up, so the cost of unpredictability is going up. That drives demand for more predictable outcomes.
The IoE business transformation
All businesses have a primary task, and face a primary risk. This risk is a factor you can influence but not control, and determines what the right primary task is. Changes in primary task can create huge anxiety. For example, Kodak was unable to transition from 'photography' to 'imaging'. Its primary risk was the rate of displacement of analogue chemistry with digital sensing and printing.
Telecoms faces a change in primary task for the IoE world. It requires going from a supply-led to a demand-led model. In the past, we constructed one-size-fits-all supply and had to find demand. The danger that kept CFOs awake at night in this model is not finding enough demand to repay the capital. This 'circuits and broadband' world sells 'pipes' with a given 'bandwidth'.
In the new model, you have to find demand and construct matching supply. The danger is of not characterizing demand well enough, or having the capability to safely engineer that supply in a 'software-defined everything' environment. This 'on demand' model sells a 'quantity of quality' at a given resilience.
The new commercial model for telecoms
This switch throws organizations into stress: roles are thrown into turmoil, as nobody is responsible for the supply/demand matching and the changing power dynamics between roles.
The business model evolves in the demand-led model. It moves away from utility 'pipes', and becomes more like a financial services 'resource trading' paradigm. This opens up many new revenue models: digital experience supply chain management; quality arbitrage; application and business continuity insurance; QoE, cost and risk portfolio management; futures and options derivatives for resources.
The underlying operating model also has to evolve. There is a massive need to collapse complexity and automate every operational task. This needs abstract models of business processes and operational performance. We need to introduce new concepts for this to happen, such as: performance invariants; predictable regions of operation for our architecture; and automated isolation of faults.
Changing mindsets and relationships
Our historical approach of over-engineering everything doesn’t scale. It just isn't viable to deliver all traffic as if it were real-time video. As Nassim Taleb documents, systems with longevity have 'optionality' baked-in, which means we need new 'antifragile' architectures. This optionality enables systems that learn under low stress and automatically adapt under high stress, a kind of 'annealing' process.
The practical upshot is that we need multiple classes of service and many resilience levels. This is a new kind of network architecture, a 'polyservice' network, with ceilings and floors on the quality of each class of service. It gives choice back to users and application developers over cost and performance trade-offs.
It’s not just the technology that has to change. The way we present our brand can (and must) change for the IoE world. Industrial users and consumer brands require trustworthiness, which means services must be predictable.
'Best effort' doesn't cut it; 'you got what you got' is not a sufficiently valuable service promise. This implies any service has to 'say what it does, and do what it says'. If telecoms companies achieve this they have earned their position as 'trustmark' brands, like Visa or Intel Inside.
To enable partnering and collaboration we need to develop clean 'interfaces'. When you 'plug in' an information appliance you want the outcome to be predictable, no matter where you do it. Where is the information 'power socket' for distributed computing appliances and applications? Thus far that interface is only defined at the electrical level, not the computational one.
Everything is change
The world is changing, and the telecommunications industry needs to change too. The two big things are always demand and supply. Demand is undergoing a paradigm shift from Information Technology to Human Technology, as we move from symbols to sensors. Supply is changing from a highly skilled numerate craft to hard science and engineering.
The Internet of Everything changes only one thing, which is everything.
Martin Geddes is a computer scientist, a scholar of technology innovation, and an entrepreneur developing new network services. His present focus is the launch of the world's first commercial "quality on demand" broadband service provider. He is also co-authoring a book "The Internet is Just a Prototype" for publication in 2017. Martin holds an MA in Mathematics and Computation from the University of Oxford.
Subscribe to the Newsletter
Join our free IoT Technical Community and receive our Newsletter.
Calendar of Events
2019 IEEE International Conference on Internet of Things and Intelligence System (IoTaIS)
5-7 November 2019
"IoT and the mmWave Frontier", an IEEE IoT Vertical and Topical Summit, co-located with IEEE RWW2020
26-27 January 2020
San Antonio, TX, USA
IEEE 6th World Forum on Internet of Things (WF-IoT)
5-9 April 2020
New Orleans, LA, USA
Call for Papers
Special Issue on Internet of Things for Smart Ocean
Submission Deadline: 1 November 2019
Special Issue on Software Engineering Research and Practices for the Internet of Things
Submission Deadline: 15 October 2019
Special Issue on Deep Reinforcement Learning for Emerging IoT Systems
Submission Deadline: 1 October 2019