Why Edge (Computing), Why Now?
With rapid advancements in sensor and communication technology and the emergence of artificial intelligence(AI) into IoT applications, the potential of this technology and its market are also experiencing fast-paced growth. According to a study conducted by Mckinsey, IoT has a total potential economic impact of $3.9 trillion to $11.1 trillion per year in 2025. On the top end, the value of this impact — including consumer surplus — would be equivalent to about 11 percent of the world economy in 2025.
However, moving away from the hype, achieving this level of impact is bound to address the technical, organizational and regulatory challenges which IoT is facing currently.
IoT and its ongoing challenges for wide adoption
For the Internet of Things to deliver its maximum economic impact, certain conditions would need to be in place and several obstacles would need to be overcome. Amongst the most important issues are interoperability, security and privacy, uncertainty in IoT related laws and public policies and technological aspects.
As for the technology, the cost of basic hardware, low-cost batteries, and low-cost data communication links are among the most important drivers for wide employment of IoT in the near future. However, for IoT users to get the most out of their data (and to use more data), there should be more and better quality (AI) tools and methods to extract insights from IoT data and the cost and availability of computing and storage of this data should continuously improve.
To elaborate further, according to an analysis by Mckinsey, only 1% of the data from an oil rig with 30,000 sensors is currently examined which is mainly for anomaly detection and control. On the other hand, optimization and prediction activities that can provide the greatest value are barely done today. This highlights the emergent need of IoT for better tools and methods to perform activities with an even higher value.
Furthermore, even in case of available high-quality tools and methods to leverage the IoT data, efficient processing and storage of this big data still remains a challenge. According to IDC, a market research firm, data generated by IoT devices will account for 10% of the world’s data by 2020, or about 44 zettabytes which highlights the importance of storage and computing infrastructure for IoT to reach a higher level of impact.
Cloud computing’s limiting factors
Although storage and computation costs play important roles in the wide adoption of IoT applications, they are not the only bottlenecks. In many IoT applications, information collected at the edge of the network (where sensors and actuators are placed) will lose its value over time — with the rate of decay depending on the nature of the information. In other words, in many IoT applications, by the time the collected data makes its way to the cloud for analysis and back to the edge (i.e where the data is collected, or where the user performs certain actions), the opportunity to act on it might be gone. Such transmission latency is too large for many use-cases such as autonomous driving, real-time computer vision applications or mission-critical systems.
Furthermore, privacy, regulatory and compliance issues, especially for classified or sensitive data, are always among top concerns for companies and the public sector to use cloud platforms. To further elaborate on this, a survey from 2015 shows that more than 50% of IoT enterprise users have serious security concerns about data leakage and control over their data.
(source: A.T. Kearney — The internet of things a new path to European prosperity)
Last but not least, connectivity, data migration and bandwidth issues associated with cloud computing are among the main cost-driving factors. For example, an offshore oil-rig that requires computational power has an option to create its own data center (facing maintenance cost and scalability challenges) or to use a cloud provider that could be located more than 5000 km away. Moreover, in the case of using a cloud provider, due to the absence of stable or economical communication to the cloud, the oil rig can only send data to the cloud a couple of times a day via a satellite connection. On the other hand, available edge sensors and devices from every oil-rig which are in order of hundreds can produce around 500 GB of data per week. On a larger scale, considering this number would upscale to millions (and soon billions) of IoT devices worldwide, even in the case of an available communication link to the cloud, transmitting such big data can burden a high communication cost and occupy unnecessary network bandwidth (if not breaking the internet infrastructure).
Edge and Fog Computing — a promising future
Edge and Fog Computing are rather new paradigms that were introduced to address the above-mentioned challenges of cloud computing infrastructure. According to Industrial Internet Consortium, Edge/Fog Computing is a decentralized computing infrastructure in which computing resources and application services can be distributed along the communication path from the data source to the cloud. Such infrastructure can bring benefits such as reduced transmission latency, better compliance, data privacy, data security, conserved network bandwidth and reduced operational cost (communication, processing and storage) for the user.
Although performing the majority of the computation tasks at the edge of the network can bring a lot of benefits, it also comes with its own limitations. For example, Edge devices are typically equipped with limited computing capacity which can translate to higher computing latency and lack of high performance for AI applications. A great portion of this limitation is addressed by scaling techniques. In the next article, I will talk about scaling approaches and how it can play a crucial role in the future of Edge and Fog computing.
DISCLAIMER: This post just reflects the author’s personal opinion, not any other organization. This is not official advice. The author is not responsible for any decisions that readers choose to make.
