Designing for IOT

Introduction

Designing Internet of Things systems goes beyond selecting the correct architecture or business planning. On this page we discuss some design principles that may inform the creation of effective IoT systems.

Design Thinking

Design thinking put humans at the centre of problem solving to ensure that the solution is desirable, useful and meaningful to the people who use it. It involves identifying a sweet spot in between requirements of the people using a product or service, the technology and the business case.

Source: Deloittes Australia

Design thinking focuses not on the business requirements of a product, but on the thinking and the process that produces these products. By starting with people, produces and services are more likely to be human-centric, holistic, and simple, delivering a desireable solution.

Typically, design thinking involves a research phase, followed by ideation and prototyping. Research involves contextual inquiry, where the end users are observed in the field. The prototyping phase involves taking the prototype back to the end users in their context to gain feedback. This process may iterate a number of times before the final design is settled on.

Design thinking involves a fostering creative mindset among participants fostering curiousity, adventure, experimentalism and optimism, reframing problems as opportunities. It often draws in related disciplines such as graphic design, UX design, industrial design and a range of artistic disciplines such as graphic design.

Uncertainty and Tacit Knowledge

A more technical design concept is the level of uncertainty that systems can tolerate while still making reliable decisions. ‘Tacit knowledge’ can be defined as knowledge that is difficult to express logically and pass on to others. This can make it difficult to define system requirements fully before building a system. In simple systems, it can be useful to use information to drive and improve system requirements through an iterative process, as shown in the diagram below.

Diagram courtesy of Ryan Messina, Messina Vision Systems

Ideally, information systems should be ‘contextless’ and ‘timeless’. Contextless means that the behaviour of the system is consistent for every environment. For example, a temperature sensor should give a reading in degrees, with an acceptable percentage of accuracy, which is unaffected by humidity, pressure, and other factors which may vary with physical position.

Timeless systems perform consistently over time. Their performance today should be identical to their performance tomorrow and in 100 years’ time. Although purely contextless and timeless systems are impossible, engineers need to analyse and define the context and time span for which the system will operate consistently. This is done using validity, accuracy and quality control.

Validity is about using the right technology for the task. If you were to measure temperature with a ruler, you should expect a very poor result, and you will get a very poor accuracy. Accuracy is defined by past performance, and cannot be used to predict future states. Quality control measures are put in place to meet expectations of future performance.

Information’s validity and accuracy is affected by the technology and sensors used, and the environment in which it is captured. Information that will be used to support decisions made by people and processes can be conceptualised as the intersection of people, process and technology as illustrated below.

Diagram courtesy of Ryan Messina, Messina Vision Systems

Key engineering challenges

A challenge is to design smart services and products that can make use of the information from potentially very large numbers of IoT connected devices. Three challenges in designing IoT data-driven services and products are:

• discovery
• integration
• analysis.

Discovery involves finding which machines and sensors are available to use. These may be legacy sensors or the property of other people or organisations. Integration involves finding methods to connect and use the data from the discovered sensors, which may be made by different vendors and produce data in heterogeneous formats. The final step is analysing the data to produce high-value information for the target product or service. This is a recurrent cycle, as the sensors on the Things are not owned by the service or product provider, and can fail, be destroyed by environmental factors or abandoned by the sensor owner. If the existing sensors are lost, discovery, integration and analysis must be repeated. The frequency of repetition depends on the volatility of the application environment.

There is also an infrastructure challenge, which is how to do all of this securely. Discovery, integration and analysis may need to be performed on the move: when either or both of the sensors and the platform that is conducting the discovery, integration and analysis are moving; and they must be able to happen in the cloud.

There are readily available open source technologies that can help in solving these fundamental challenges. These include:

• OpenIoT for sensor integration
• Semantic Sensor Networks (SSN) ontology for sensor discovery.

Sources:

The information on this page has been sourced primarily from the following:

• A webinar titled 'How Machine Vision Helps Realise the Smart City Concept' by Ryan Messina, Director and System Engineer, Messina Vision Systems delivered to this community on 4 July 2017
• A webinar titled IoT application development with open data-driven computing platforms by Prof Dimitrios Georgakopoulos, Swinburne University of Technology
• A webinar titled Design Thinking for the Internet of Things by Dr Lauren Tan, Director-Design for Business and Betrand Marcau from Deloitte Touche Tohmatsu