Digital transformation is on the rise as businesses are competing to set an industry benchmark. Internet of Things is one of the key digital transformation technologies that can give the edge to these businesses in order to gain the technological advancements they seek.

To embolden economic development and higher quality of life, “Smart City” is remarkable as well as an exceptional digital model for innovations. The concept of smart cities is a combined package of smart home systems, smart transport, intelligent buildings, smart traffic lights, smart waste collection, and everything else that you think as something that can use or provide valuable data for smart city management.

On the one hand, the new Internet of Things (IoT) applications are empowering smart city activities around the world, on the other hand, it is also increasing cost and latency as all of this data is sent to the cloud or centralized data centers for further analysis and processing. In such a scenario, IoT deployments need to process information and decision-making closer to the source of the data (at the edge), thus cutting the costs and time associated with cloud data transfers.

The concept of the sustainable smart city comes with connected everything — people, devices, and processes. It gives a smarter digital world where everything is connected, communicating, and improving the standard of living. This idea of a smarter city introduces smart buildings, traffic management, which can reduce traffic congestion, water management, waste management, etc.

Role of edge intelligence in the sustainable smart city development

Apparently, the data collected from different smart city components are huge. Hence, data deluge is a challenge you need to think about while planning for a sustainable smart city since it can be difficult and expensive to manage. To address this concern, smart city initiatives can adopt edge analytics, which collects the data and analyzes it. In edge analytics, automated analytical computation is performed on data received from sensors, network switches, or other devices instead of sending back to a centralized data store. By running data through an analytics algorithm at the edge, companies can set parameters to decide what information is worth sending to the cloud for later use. This decreases the latency in the decision-making process for connected devices.

Why edge intelligence for smart city?

As mentioned above, edge analytics is an approach to collect the data from sensors, switches, or other devices and analyze this data. Edge intelligence is nothing but edge computing with machine learning. Edge intelligence brings data pre-processing and decision-making capabilities closer to the data source, which reduces data deluge and delays in communication. Thus, the time and cost can be saved with the edge intelligence structure, which would be a key performance indicator for smart city management.

Advantages of Edge Intelligence

• Near real-time decision making
• Lower latency
• Reduced communication cost
• Autonomy
• Fully distributed computing model and local identity
• Enhanced quality of data
  • Reduced data volume
  • Pre-processed data so that only decisions or alarms can be forwarded to the cloud servers, rather than raw data

“By 2020, the spend on edge infrastructure will reach up to 18% of the total IoT infrastructure spend.” – IDC

Different forms of edge computation

Edge Analytics

Edge analytics gives analytics of the data at the edge of a network either at or close to a sensor, a network switch, or some other connected device. Whether it is about rule-based decisions or complex event processing design responsible to proactively handle incidents/situations, all of that can be handled by edge analytics. With smart sensors and connected devices, edge analytics requires hardware and software platforms for storing, preparing the data, and training & processing of the algorithms. The capacity of processing and storing data at the edge also plays a key role.

“The edge analytics market is estimated to reach USD 7.96 billion by 2021, at a Compound Annual Growth Rate (CAGR) of 32.6% from 2016 to 2021.”

-MarketsandMarkets

Machine Learning at the edge

Machine learning is an ability of the system to automatically learn and improve its operations without any human interference. For example, in the case of self-driving cars, machine learning applications are trained locally in the car itself (at the edge) to cut back on bandwidth and latency to process data. Not to mention, the life-safety factor required; the ability for these vehicles to process data instantly and make decisions based on the current road conditions is critical and can be life-saving. This is going to help in identifying speeding vehicles and cyclist movements, improving traffic flow, enhancing pedestrian safety, and optimizing parking.

Pattern Recognition at the edge

For the pattern recognition, machines are trained exactly the way human brain thinks and recognizes the patterns by analyzing various aspects of the object. Pattern recognition in AI is where machines are trained to recognize the required images based on a particular pattern. Learning and predicting traffic and parking patterns and other logistical data at the edge can be possible. For instance, if the system spots some event happening at an intersection A, it will predict the real-time impact at other intersections.

Design requirements for moving towards the edge

There are various properties and requirements that need to be considered when adopting edge intelligence, such as credibility and trust, autonomy, machine learning capabilities, self-organization, self-configuration, self-discovery, self-learning & self-adapting, policy-driven operations, mesh capabilities, resiliency, and semantic interoperability.

Any IoT platform to collect and analyze OT data should be extremely adaptable in the manner in which it accomplishes this. OT data can be acquired from an assortment of sources as mentioned below:

• from sensors using a specified protocol,
• from control systems like SCADA, DCS, and other proprietary systems that control hardware,
• from operations software like ERP, data historians, process management systems or any custom data logging or monitoring systems.

Applications of edge analytics

Smart Parking Solution

Every single day people waste a great deal of their time searching empty parking slots. This wastage of time can be controlled if people know exactly which parking slot is available or empty and how to proceed for that empty slot.

Smart parking is a typical IoT application that can reduce parking queues and quickly find empty space by identifying the real-time availability of free parking slots. The system utilizes various concepts and image detection systems along with the already-present surveillance cameras. This edge intelligence visual parking slot occupancy detection system can be developed through Convolutional Neural Network (CNN) techniques, that can be designed and deployed for smart cameras.

Home Automation

Home Automation is another application of the Internet of Things through which you can improve home safety. A truly intelligent smart home is a multi-layer system, which requires little to no management on a user’s part and is capable of making decisions based on historical and real-time data. Thus, the home automation system should be able to identify significant user actions, assess the probability of events, and issue appropriate commands to other devices within the network.

The Nest Cam IQ indoor security camera, for example, uses the image and video recognition as well as machine learning algorithms at the edge to monitor motion or to verify family members’ faces, allowing smarter security breach detection and fewer false alarms.

Traffic Management and Smart Transportation

Traffic management is an ideal application for edge computing technologies. By deploying compute intelligence locally, on the physical traffic, noisy data can be reduced at the edge. This significantly reduces the amount of data that needs to be transmitted over the network, thus reducing operating and storage costs.

Smart transportation provides live streaming traffic information powered by machine learning and edge computing. It reduces traffic accidents with connected infrastructure, data analytics, and machine learning that can optimize traffic systems and identify high-accident intersections.

Machine learning, when deployed on the edge devices, will enable connected traffic systems to engage with and react to both manned and autonomous vehicles. Machine learning tools collect traffic data from the Internet of Things (IoT) sensors that are embedded in the roads and in the traffic lights, historical surveys, radar images, etc. It studies the traffic pattern and figures out when heavy traffic begins and ends. Then the intelligent traffic system dynamically adjusts the signal timing of traffic lights based on the learning. Machine Learning aggregates and analyzes real-time data to control traffic lights and the overall transportation system.

To sum this up, the benefits of edge intelligence with machine learning reduce the data deluge and minimize the delay in communication, which ultimately reduces the cost.

eInfochips is a Microsoft Gold Cloud Productivity  and an AWS Advanced Consulting Partner. eInfochips helps clients in implementing a highly scalable, reliable, and cost-efficient infrastructure with edge and fog computing capabilities to help you with smart city implementation. To know more about eInfochips, get in touch with us.