How LoRaWAN and Edge AI are revolutionizing the Internet of Things

The integration of artificial intelligence (AI) and the Internet of Things (IoT) in practical applications has formed AIoT (intelligent Internet of Things), which is the intelligent interconnection of all things. At present, AIoT has become the mainstream form of future technology recognized by the industry. The "2020 China Intelligent Internet of Things (AIoT) White Paper" recently released by iResearch Consulting predicts that by 2025, the number of IoT connections in China will be nearly 20 billion. The interaction and data analysis needs generated by massive connections will promote a deeper integration of IoT and AI. On the other hand, intelligent IoT systems expose potential problems in cloud computing. The increase in intelligence and automation inevitably leads to unpredictable delays in applications with outstanding performance and security.

Today, two major challenges threaten the multiplying number of connected devices: the performance of edge devices for long-range communications, and the battery life of off-grid IoT applications.

The transmission of raw data is very power-hungry for any device. Traditional cellular wide area networks (WANs) consume a lot of power and are therefore not suitable for battery-powered IoT devices. IoT applications LoRaWAN (Long Range, Wide Area Network) is one of the preferred communication protocols in IoT applications and can address how artificial intelligence is changing the IoT architecture through edge applications.

Why use LoRaWAN and edge AI?

With the proliferation of smart devices, both core network domains and end devices face communication challenges such as congestion, security, service latency, data privacy and lack of interoperability.

For the network domain, most of the challenges come from over-reliance on cloud computing. When sending data to the cloud, more energy consumption, bandwidth, storage, and latency are incurred, leading to higher costs. Fog computing or edge computing can reduce costs and improve efficiency.

When wireless technology is used for data transmission, communication barriers in the end device appear. In the Internet of Things, the advantage of Bluetooth and other wireless standard technologies is low power consumption, but limited coverage is a major obstacle, especially for smart city services. In this case, low-power wide area networks (LPWAN) have become a reliable alternative between long-range cellular and short-range operating technologies.

LPWAN is a low-power and wider-coverage communication physical layer that operates in the Sub-GHz unlicensed radio frequency band. LPWAN is a standard protocol that is effective for the link and network layers, providing variable data rates and increasing the possibility of exchanging throughput for link robustness, coverage or energy consumption. Both organizations and individuals can deploy LPWAN networks.

LPWAN and Fog Computing Architectures Close to the Edge

In terms of intelligence and data processing, edge computing and fog computing look similar. However, the key difference between them lies in where the computing and intelligence occurs.

A fog computing environment places intelligent processing on a local area network (LAN) that transmits data from endpoints to a gateway. Edge computing, on the other hand, places processing power and intelligence in devices such as embedded automation controllers.

These devices can run algorithms to produce edge intelligence—a product of AI and edge computing.

Advantages of using LPWAN for edge computing

Reduced data transmission: Edge computing reduces the amount of data transmitted and stored in the cloud. Another advantage is that placing computing power at the edge of the network minimizes latency and costs while alleviating bandwidth requirements.

Reduced latency: Edge computing minimizes the time between data transmission, processing, and actions taken based on insights from the process. In addition, the speed of analysis and event processing is increased at a lower cost, and the signal-to-noise ratio is reduced. Due to the closer location to the end user, the bandwidth and power consumption of the core network and connected devices can be reduced. Therefore, edge computing provides low-latency capabilities through real-time services, which is essential for smart cities, vehicle-to-vehicle communications and other applications that require latency less than tens of milliseconds. This is lower than the latency of mainstream cloud services.

Enhanced security: Most users consider data security and privacy as their top concerns, mainly because these factors pose a security threat to smart city-related applications. Security must be divided into three layers: user privacy, data security, and network connectivity. Edge computing addresses the challenges of IoT security through measures such as credential upgrades and security checks on multiple physical devices.

Expanded applications: LPWAN and edge devices are ubiquitous in healthcare monitoring, for example, to detect patient falls. Edge devices can improve accuracy and adaptability in filtering data for real-time processing. In traditional systems, the raw data sequence is transmitted in the cloud, so the delay of alarm increases. Edge systems reduce the computational work on sensor nodes by shifting the heavy computational load from sensor nodes to edge gateways.

How to use edge artificial intelligence to accelerate the implementation of application scenarios

While the model building and training phases for edge devices consume significant resources and add additional complexity, there are high-quality options on the market that offer customization and reduced complexity.

Avnet's SmartEdge Agile device can simplify and significantly reduce this complexity. SmartEdge Agile is an edge computing device equipped with various types of sensors. Brainium is used to build and train models. The device has LPWAN connectivity to establish a fog computing architecture and uses a gateway to connect to Brainium. Avnet's SmartEdge Industrial IoT Gateway connects Brainium and the cloud securely and seamlessly.