Many technologies must deliver for IoT to deliver
Internet of Things or IoT aims to connect interrelated computing, mechanical and digital machines to a common platform. The devices will connect to each other as well as to the humans and will collect information about their environments by sensors. The number of such devices could touch 20 billion by 2020 as per Gartner. As per International Telecommunication Union number of humans using internet is 3.2 billion. In the emerging digital world, connected devices or things would overwhelm the humans.
The impact of IoT would be unprecedented. Home appliances could be remote controlled and warnings of faults would be available on time. In industries, control of various devices in the factories will shift from factory floor to the control rooms. Autonomous cars will predictive maintenance needs would become norm and their safety and fuel efficiency will improve. This will apply to other modes of transport too. Smart meters will synchronize demand supply curves and reduce distribution losses. Smart cities would emerge which would have improved services, less traffic congestion, better conservation of water and energy, and improved quality of life. The chances of predicting onset of certain diseases and successfully managing them after patient catches them will improve. The list is endless.
The key characteristic of IoT is the amount of data. IDC’s Digital Universe study predicts the world’s data will rise from 4.4 Zettabytes (10^21 bytes) in 2013 to 44 by 2020, 10% of this will be from the IoT. By 2025, it would touch 180 Zettabytes. In some cases, the data generated by individual device would be enormous e.g. self driving car from Google generates 2 Petabytes (10^15 bytes) of data in a year. Besides, the proportion of data that can be analyzed will also increase from 22% in 2013 to 35% in 2020, driven by IoT.
Cloud is often considered as an innovation to business models as it allows companies to outsource their storage and computing needs, while they focus on their core competencies. However, with IoT it is technology enabler. If all IoT devices need to store and process their own data this will make IT a significant part of such devices, forcing change in assembly line processes as well significantly increasing their maintenance costs. One way to reduce these entry barriers for adoption of IoT is to use cloud for storing and computing needs. The cloud infrastructure will allow for analysis besides efficient storage. However current capacities of cloud providers will be challenged. As per the website www.statista.com, the capacity of data centres offering public cloud was 465 Exabytes(10^18 bytes) in 2017. The scale up needed to accommodate IoT is massive. Currently, hard disks and magnetic tapes capable of storing data in Terabytes (10^12 bytes) are available. However, as data will increase exponentially, cloud providers will need to work on other storage technologies e.g DNA, HVD etc
5G telecom network
This data will generate network traffic. IoT cannot deliver its promise on existing networks and it needs a a network with much higher speed, low latency and consistent performance. Rollout of 5G network is critical for widespread adoption of IoT. The proposed speed is 20 Gbits/sec as compared to 100 Mbits/sec for 4G networks. This would be accompanied by limits on user experienced speeds i.e per-user download speed of 100Mbps and upload speed of 50Mbps rather than just the theoretical maximum, lower latency at 4 ms, support for higher device densities at 1 million connected devices per square kilometer etc. These parameters are specially targeted for IoT. Conversely, IoT is an important use case of 5G networks to justify the investments. Some applications of IoT e.g. driverless vehicles not only need to transfer huge amount of data but also with minimal latency as delays could be dangerous, for a car needs to make real time decisions to avoid accidents. The latency of 4G was at least an order of magnitude worse and hence incapable of supporting driverless cars.
This latency issue will also warrant change in cloud paradigm. One way to analyze IoT data is near the devices that produce it and act on that data, referred to as Fog computing. The fog nodes, can be deployed anywhere with a network connection e.g. on a factory floor, on top of a power pole, , in a vehicle, or on an oil rig. This reduces bandwidth pressure on the long distance network and can handle cases where latency is critical. The percentage of devices that need computing power near to themselves will increase in future.
Analytics and Machine Learning
Data needs analysis. Big data algorithms hosted in cloud are needed for this analysis as data would come in various formats, huge volume and needs to be processed at fast speeds so that it could to be useful. But one of the proclaimed benefits of IoT is in predictions. Here Machine Learning (ML) could play a critical role. ML is defined as the ability of a machine to vary the outcome of a situation or behavior based on knowledge or observation which is essential for IoT solutions. ML could allow useful patterns, correlations and anomalies to be searched. It can also predict unknown outcomes. Different ML algorithms will need to be “trained’ for different use case of IoT
Security and Privacy
Security and privacy issues are significant hurdles to IoT penetration. The IoT devices will be always on and connected bringing new challenges to security. They will be additional challenges due to number of devices involved for which security measures will need constant upgrades.
In 2013, a hacker got access to credit card information of customers of Target by using network credentials taken from a heating, ventilation, and air conditioning vendor. 2015, 2 researchers demonstrated a wireless hack into Jeep Cherokee, first taking control of the entertainment system and windshield wipers, and then disabling the accelerator.
There is additional issue of privacy as huge amount of personal data would be captured by IoT devices which could be hacked or intentionally used for business purposes.
IoT devices will need battery power to remain always connected. New wireless standards such as Bluetooth® Smart (also known as “BLE”) or zigbee reduce battery consumption and allow coin cell batteries to be used. But the sheer numbers have led to exploration of an alternative approach called Energy Harvesting (EH). Also called power harvesting it is a technology that aims at collecting energy from ambient external sources such as heat, light, vibrations, radio waves etc. It produces very low power levels on the order of several microwatts (10-6W) to milliwatts (10-3W) but that is enough for sensors that are battery optimized.
IoT will impact every aspect of our lives, including our homes, offices, cars and even our bodies. It will bring structural changes in global economy and as per McKinsey may generate upto $6.2 trillion in value by 2025. However many technologies need to deliver their expectations for IoT to deliver its promise