DS in the Real World
5G
Why Data Scientists and ML/AI Engineers should care
5G is a hot topic in technology circles, political conversations and in the news. Interestingly enough, it was also a hot topic at Davos, according to McKinsey, for the high potential of disruption, opportunities and controversy around its implementation. I think 5G will create interesting and fun opportunities for data scientists.
The Basics:
5G networks are the next generation of mobile internet connectivity, offering faster speeds and more reliable connections on smartphones and other devices than ever before.
5G combines cutting-edge network technology and the latest high-speed devices. The expectation is that 5G will deliver data with less than a millisecond of delay (compared to about 70 ms on today’s 4G networks) and bring peak download speeds of 20 gigabits per second (compared to 1 Gb/s on 4G) to users.
5 technologies make 5G possible
In simple terms, the 5 technologies powering 5G are:
1. Millimeter waves: short waves in the higher end (30–30GHz) of radio communications broadcast frequencies. This higher end of the spectrum has, until now, been used mainly for satellites and radar systems. It has not been broadly used for mobile communications. The advantage of using millimeter waves is more information can be transmitted faster. The disadvantage is these waves can only travel short distances and cannot penetrate structures.
2. Small cells: portable miniature base stations that require minimal power to operate and can be placed on poles and other structures in a city every 250 meters or so. Having more stations means the frequencies that one station uses to connect with devices in one area can be reused by another station in a different area to serve another customer.
3. Massive MIMO: MIMO, which stands for multiple-input multiple-output, describes wireless systems that use two or more transmitters and receivers to send and receive more data at once. Massive MIMO takes this concept to a new level by featuring dozens of antennas on a single array. Today’s 4G base stations have a dozen ports for antennas that handle all cellular traffic: eight for transmitters and four for receivers. 5G base stations can support about a hundred ports, which means many more antennas can fit on a single array. That capability means a base station could send and receive signals from many more users at once, increasing the capacity of mobile networks by a factor of 22 or greater.
4. Beamforming: The primary challenge for massive MIMO is to reduce interference while transmitting more information from many more antennas at once. And this is where beamforming comes in. Beamforming is a traffic-signaling system for cellular base stations that identifies the most efficient data-delivery route to a particular user, and it reduces interference for nearby users in the process. Beamforming can help by focusing a signal in a concentrated beam that points only in the direction of a user, rather than broadcasting in many directions at once. This approach can strengthen the signal’s chances of arriving intact and reduce interference for everyone else.
5. Full Duplex: Today’s base stations and cellphones rely on transceivers that must take turns if transmitting and receiving information over the same frequency, or operate on different frequencies if a user wishes to transmit and receive information at the same time. With 5G, a transceiver will be able to transmit and receive data at the same time, on the same frequency. This technology is known as full duplex, and it could double the capacity of wireless networks at their most fundamental physical layer.
What 5G means for data science
So, what does 5G technology mean for data scientists? Well, 5G will have the ability to support up to a million concurrent edge devices per square kilometer. That scale will enable businesses to collect vast amounts of data continuously from mobile phones, sensors, thermostats, and other 5G-equipped devices. The amount of data, and the speed at which data will be available, creates multiple opportunities for data scientists in different industries, if not all.
Let’s begin with the some of the industry applications:
● Some of the more popular use cases are in autonomous vehicles. While today most of autonomous driving is self-contained, with the advent of 5G, more complex algorithms for object detection, object recognition and classification, localization and prediction of movement will make use of real time data transmitted from vehicle to vehicle and traffic control. Anil Gupta provides a more detailed review of some ML algorithms used in autonomous vehicles in his article. “Machine Learning Algorithms in autonomous driving.”
● But, there are other less touted and equally interesting use cases across many different industries. Let’s take a look at agriculture:
○ Precision farming: real-time and historical data, along with ML algorithms, can more broadly be used to make specific decisions for small areas of application for pesticides, fertilizers and other chemicals to increase crop yields and reduce environmental impact.
○ In animal identification and health monitoring, real time streaming of images can be used to assess the health of a particular animal to prevent the spread of diseases.
○ Crop monitoring systems would help many farmers to produce accurate diagnostics of individual areas or even individual plants growing in the field, giving them the ability to control their harvest and its potential. For a fun and detailed view of the multiple applications of ML in agriculture, read this article by Oleksii Kharkovyna.
● Let’s continue with healthcare, where despite the controversies regarding privacy and security which keep healthcare at the forefront of AI news, wearables and sensors are being developed for Remote health monitoring. In this application, information about a patient’s health can be continuously relayed to their doctors. This helps give doctors the ability to anticipate complications after surgery or monitor blood sugar levels for preventative care, meaning the potential to save lives is greater than ever before. Another application in healthcare includes real-time video streaming for remote medicine. At the Mobile World Congress 2019 in Barcelona, Vodafone exhibited two new health telecom technologies it developed leveraging 5G: in one, paramedics stream real-time video to get remote support from specialists while transporting patients. In the other, the extremely low latency allows specialist clinicians to guide surgeons in real time from remote locations to perform remote surgery using robotic arms.
Opportunities in manufacturing and security include:
● Automated industrial processes and inventory optimization. For instance, advanced sensors will pick up levels of performance data about the heavy machinery running inside plants — which will be transmitted at superfast speeds between those machines and central control systems for precision manufacturing and real-time adjustments.
● In the area of maintenance, 5G could take preventative maintenance to a whole new level; real-time performance data from the machines on the plant floor can be coupled with external data to identify potential breakdowns and schedule maintenance crews and spare parts in a highly automated fashion reducing down time even further.
In cybersecurity, 5G could be leveraged for network threat detection (more on this here).
Imagine what providing contextual awareness to wearables and virtual assistants will be like: they learn your moving patterns and remind you that usually you are at Battery and Pine at 5:20 (your bus stop). Or learn your best running pace and suggest when you need to pick up your speed to lose weight.
DS and ML to Build a Better Network
The other opportunity for data science and ML in the world of 5G communications is the network itself.
5G networks need to be much smarter than previous systems, as they’re juggling many more, smaller cells that can change size and shape. There are also the multiple technologies that enable 5G mentioned earlier that need to be managed: massive MIMO, full duplex and beamforming. Add to this the wide spectrum of available frequencies which will enable the flow of streaming data from all types of connected devices using different ranges of the spectrum, and it becomes quickly apparent that managing this complexity at scale will require the use of machine learning (ML).
To fulfill the promise of 5G, networks will have to be self-healing, self-managing, self-securing, self-repairing and self-optimizing. Embedding ML and other AI models to automate application-level traffic routing, quality-of-service assurance, performance management, root-cause analysis, and other operational tasks will be required. It will be impossible to manage all these tasks effectively and at scale otherwise.
Two enabling elements according to Vikram Pulakhandam (see ref [2]) are:
● “Anomaly detection will be one of the key ingredients to ensure the underlying network and customer support systems can self-heal and self-organize to keep pace with the high level of service requirements (low latency, high reliability) that are required to support mission-critical applications”
● “Forecasting demand and behavior will be key to enable service providers to automate the creation of network slices to support new usage patterns. Forecasting user demand will ensure sufficient network resources are provisioned, created and scaled ahead of usage. Taking service creation to service consumption in order to meet changing consumer demands will be the killer value proposition for 5G service providers.”
So, as we just saw, data science, ML and AI thrive on massive amounts of data, and 5G will create massive amounts of data. 5G provides the high-speed network to move the data that AI needs to succeed, and AI potentially has the ability to handle the complexity of 5G.
The combination stands to enable a new generation of applications and therefore open new opportunities for data scientists and AI/ML developers where data availability/speed was a limitation. I am eager to read more and hear from you about other opportunities you see potentially rising with the deployment of 5G networks and high-speed devices. We can also discuss the impacts on human health, national security and economy, the topics that make the headlines.
For more on 5G, you can check out this video recorded at a Watson Bootcamp, and featuring Dr. Faisal Akkawi, Executive Director, Information Systems programs, School of Profession Studies, Northwestern University.
References
[1] Amy Nordrum, Kristen Clark and IEEE Spectrum Staff, Everything You Need to Know About 5G, 27 Jan 2017 ,IEEE Spectrum
[2]Vikram Pulakhandam, 5G is Rolling Out: Here’s How Cognitive Analytics Will Take Part in the Revolution, Anodot
