With each new generation of wireless technology, development opportunities for mobile solutions have rapidly advanced. The introduction of the iPhone 3G in 2008 changed the way users interact with their devices, accelerating the adoption of the third generation of wireless technology. 3G offered data transfer speeds that enabled internet browsing, email, content streaming and powerful apps to become mainstays of the smartphone experience. When 4G came to Canada in 2011, it transformed the way we live and work, with handheld devices running robust applications, exchanging massive amounts of data and becoming extensions of IT environments.

The fifth generation of mobile networks, 5G offers faster speeds, lower latency and the ability to connect even more devices than previous generations. In this article, we will examine several key developer considerations as 5G networks continue to expand:

• Opportunities of 5G for developers: high bandwidth, low latency and edge computing;
• Use cases: applications and services well-supported by 5G;
• Compute platforms: target devices and servers needed to create your applications and services;
• Accessing the radio access network (RAN): considerations for the wireless network that connects devices, servers and cloud platforms;
• Availability: expectations and challenges of a 5G network.

5G opportunities

5G delivers tangible opportunities in three main areas:

1. High bandwidth

Bandwidth represents the rate of data transfer across a network, often measured in megabits per second (Mbps). 5G technology deployments will have theoretical peak download speeds that are up to 100x faster than 4G LTE¹.The bandwidth available to any one specific device will depend on:

• The number of devices on the network;
• The strength and quality of the wireless signal available to a specific device;
• The quality-of-service configurations which can be used to limit speeds to manage the priority of access among many devices

Bandwidth is important to support the quality of applications and services that have high payloads (the need for a lot of data traffic). For example, if the payload is a digitally encoded video stream, or a point cloud from a light detection and ranging sensor (LiDAR), data rates generated from a single device might be on the order of 15 to 50 Mbps. When many devices are applied in a single application or service, the importance of a reliable network to flawlessly send and receive these data streams must be considered.

2. Low latency

Latency refers to the delay between sending and receiving information, with 4G usually offering below 50 milliseconds². and 5G going down to just one millisecond under ideal conditions³.

A 5G networked device, such as a wheeled robot, contains a collection of sensors (to provide its location, velocity and video of its surroundings) and motors to control speed, direction, and to perform actions (such as the ability to reach and collect items on a shelf). There are higher-order algorithms, often supported by artificial intelligence or machine learning engines, which may be applied to provide intelligent navigation, identification of items captured on video, or to avoid collisions with other robots or physical obstructions.

These algorithms may perform better on higher-powered compute platforms on edge or cloud platform servers, rather than on individual microprocessors that are available in robotic systems.

With low latency, the sensor data can be provided by the robotic systems to a server within a premises or in the cloud. The algorithm may generate and send navigation instructions back to the robot to satisfy navigation, safety or decisions on which items to collect. These instructions must be provided in real time to the robot, as if they were generated locally by a processor in the robot. Low latency provides this real-time feedback.

3. Edge computing

Edge computing brings computation and data storage closer to the sources of data. When devices communicate with servers over a low-latency, high-bandwidth wireless network, those servers can either be on-premises compute platforms, compute platforms close to the premises, platforms administered by a mobile network operator (such as Bell Public MEC on AWS Wavelength) or a general-purpose regional cloud platform provided by a cloud platform provider.

We consider either the on-premises server or the compute platform close to the premises to be an edge server. Typically, these are high-performance instances that provide computing, graphical processing, memory, storage and advanced algorithm features. Due to the low latency and high bandwidth between devices and their sensors (or actuators, such as motors), applications and services may be distributed between devices and edge servers based on where workloads may be best executed, or where data may be best aggregated for decisions and analysis.

Compelling use cases

Compelling use cases are already being developed and implemented in a wide range of applications. Solutions for smart cities, utilities and infrastructure management, remote healthcare, augmented reality and robotics are reaching maturity. There are also compelling use cases in supply chain management offering the ability to track, monitor and optimize the movement of goods and the systems that transport them in real time. The benefits of these use cases can include improvements in safety, sustainability, maintenance, resource allocation, risk mitigation and security.

A wireless network featuring high bandwidth, low latency and tightly-integrated edge computing provides solution architects and developers with the opportunity to distribute the computational, data aggregation and functional workloads of a solution where each workload is best suited. This distributed approach is supported by well-received, industry-standard foundational technologies for integration and workload balancing.

The automated guided vehicle (AGV) and advanced driver assistance use cases are good examples where devices, sensors, actuators and edge computing over 5G can offer compelling solutions for road safety and transportation efficiency.

Today’s vehicles are intelligent devices. Their operational state data (velocity, acceleration, direction, position, braking, steering wheel angle rotation and data from externally focused sensors such as cameras and LiDAR) is continually collected, monitored and processed to provide safety, driver assistance and vehicle diagnostic information.

If this information is aggregated from vehicles approaching an intersection or even at the level of a municipality, the data can be processed and acted upon in ways that will lead to significant opportunities to enhance the safety and management of vehicles, pedestrians and traffic flows. If we add the condition of road infrastructure such as traffic lights, traffic cameras, digital signage and the real-time state of pedestrians and cyclists, developers will be able to create applications that provide real-time traffic management (like balancing the timing of traffic lights), notifications to drivers, updated messaging to digital signage and safety warnings to pedestrians and cyclists. As vehicles approach fully autonomous operation, the messaging can affect the actuators on the vehicles (steering, braking, acceleration) based not only on what the vehicle observes directly through its sensors, but based on the state of surrounding vehicles, road infrastructure and variable obstructions (such as pedestrians).

The real-time requirement of multi-vehicle data collection, aggregation, processing and feedback can be facilitated by 5G connectivity and edge computing. The solution would be highly distributed, with large datasets and the higher-order computational power that is required to understand the state of all participating vehicles.

This use case illustrates how developers have explored the ways distributed computing in a capable 5G network can be leveraged successfully. Developers can now explore future applications and use cases – in spaces that must wirelessly connect devices, sensors, actuators and compute platforms. Developers may envision how their applications or services may be constructed in a distributed manner and rely on 5G infrastructure and edge computing as the means to host it.

Compute platforms

In any mobile computing environment, there are a number of platforms where selected components of full applications and services may be executed. The computing platforms may be associated with devices that are highly integrated with 5G networks such as end devices and edge devices (such as a router). Other platforms may be cloud-based or on-premises – platforms that may be highly capable computationally and which may host large data sets. While these platforms are networked to a 5G environment, they may not realize the real-time low-latency properties we’ve been discussing. They can contribute within a full application or service architecture, but are best utilized when low latency isn’t required.

Typically, end devices such as automated guided vehicles (AGVs) are the originators of data collection. AGVs are best optimized for sensor integration, data collection, localized decision making as well as for providing data to higher-order platforms for aggregation, system-level coordination and decision-making. Often these devices have significant compute engines to accomplish the localized computational and networking functions. Sometimes they may only have sufficient computational power for sensor data integration and telemetry of the data to higher order-platforms.

End devices often provide developer-accessible interfaces (network APIs) so that their operation may be commanded (such as notifying a robot to stop immediately when a safety hazard is determined) or to access their localized sensor data (video streams, point clouds, GPS coordinates, etc.). The ability for developers to add functional features that are not supported by their network APIs is often limited. This provides another rationale for a higher-order compute platform where functional features or advanced algorithms may be remotely supported with an end device by using the available network API in near real time (low latency) with large data sets (high bandwidth).

The edge device, highly integrated within the 5G radio access network, can be located where other end devices are integrated and directly managed by end customers. These are often general-purpose computing and graphical platforms where developers can build their applications and solutions with open-source operating systems and functional software distributions.

Edge devices may also be localized extensions of cloud computing platforms. An example is Bell Public MEC with AWS Wavelength, which offers both the tight 5G network integration of an edge platform as well as the availability of familiar products such as Elastic Compute Cloud and machine learning engines. This extends the feature set of tools for developers – features widely applied by cloud computing developers today.

Edge device-based applications can (and should) be virtualized and containerized to provide the flexibility for porting and distributing their solutions, as those solutions need to consider scalability and the flexibility to be hosted on emerging next-generation platforms.

Accessing the radio access network

Developers must consider that not all devices and on-premises edge platforms have integrated 5G modems – especially for device and on-premises edge platform connectivity to 5G. To bridge the native network capability of devices to the 5G RAN it will be necessary to include 5G routers or adapters.

It will also be necessary to evaluate whether the introduction of routers or adapters affects latency or bandwidth. Developers and solution architects should test end devices, edge devices and proposed 5G routers to ensure they provide the required bandwidth and latency characteristics of the solution.

Availability

Beyond the accessibility of devices, it is essential to understand the 5G network coverage map based on all locations where end devices will likely be positioned. The 5G solution is more than the software architecture of supporting how components are distributed among end devices, edge and cloud platforms. Comprehensive network coverage planning is essential.

Solutions must also behave coherently when devices potentially lose connectivity, temporarily or permanently. When a solution requires a specific end device’s data set or operation, failover and redundancy scenarios need to be considered.

Private vs. public mobile networks

From a developer’s perspective, public and private networks provide similar connectivity approaches. However, some use cases will require the specific capabilities of a private network. The key differences between public and private networks relate to priority access and isolation. Typically, public networks available from mobile network operators offer equal access rights to all users. A private network offers greater control and can be configured to allow different levels of priority access when certain network activities are deemed more business-critical than others.

For use cases that require a private network, initial testing of distributed end-device, edge-device and cloud platforms may be achieved by using a public network (defined as an environment generally accessible by a mobile network operator’s individual subscribers). This will provide a quick start to validating the expectations of a high-bandwidth, low-latency environment. If the solution is meant to function within a private-network setting, final solution testing and deployments will need to be performed in the specific network that will service the area in question.

Next steps

While use cases and development opportunities abound, there are likely to be new applications for 5G that have not yet been envisioned, and new industries that will be created. The range of computing platforms, the radio access network and the availability of those networks will also continue to expand. To stay up to date, you can continue your 5G journey by reading Exploring telco 5G edge computing.