Latest from Network Transformation/Edge Compute/IoT/URLLC/Automation/M2M

66cddf9e24f0193bd7f8e665 Arelion Denmark010 Digital

Solving the Transformation Puzzle

Oct. 28, 2024
Why AI is just one piece of network evolution.

Why AI is just one piece of network evolution.

Network evolution is unpredictable, but we can approximate its future by examining the past and present. Telecom has had a wild ride from its humble beginnings rooted in legacy protocols.

While some networking technologies have been phased out, dense wavelength-division multiplexing (DWDM), Ethernet, and others have stood the test of time. The industry then experienced the rise of cloud services amid hyperscaler growth, challenging how operators build their networks.

Since AI relies on massive amounts of high-quality data, telecom operators must collapse their internal data siloes to create a unified source of truth. Despite its transformative power, AI is only as good as the data you feed it.

Now, artificial intelligence (AI) will transform networks and network operations again, but it may only be one piece of the bigger puzzle. So, where do networks go from here? Let’s explore networking’s early development to contextualize its evolution amid these emerging applications.

From Gas Station Bathrooms to Global Infrastructure

Early networking infrastructure was placed in unconventional locations, with some network operators’ amplifier sites even placed in gas station bathrooms to save both cost and installation time. While this seems strange, these sites can function optimally for years. As connectivity expands globally, operators will continue to choose creative infrastructure locations according to companies’ business needs.

Various legacy protocols dominated the early networking landscape, including X.25 and frame relay. As an early packet-switched networking protocol, X.25 was designed to maximize the reliability of communications but was plagued by inefficient error checking and correction mechanisms.

Frame relay replaced X.25 as a high-performance wide area network (WAN) protocol designed to maximize cost efficiency while reducing the complexities of error correction. However, scalability was frame relay’s Achilles heel, with its fixed-size virtual circuits hindering the management of varying traffic loads.

Multiprotocol Label Switching (MPLS) emerged as a dynamic alternative, offering improved flexibility and performance as networks and internet traffic scaled up in the late 1990s. Throughout these changes, network operators worked to maximize efficiency, cost savings, scalability, capacity and more.

No matter how the network evolves to accommodate AI’s requirements, operators will integrate new technologies or leverage innovations in existing technologies to enhance these same networking qualities that have proved vital since the internet’s early days.

Tomorrow’s Puzzle Pieces: Where Networks May Head Next

AI and Optical Innovation: A Symbiotic Dynamic

AI has captured the technology industry’s imagination and wallet, and networking is no exception. AI will catalyze a two-pronged transformation, where operators will build their networks to maximize automation, high-capacity bandwidth, cost savings, speed and reliability to serve AI’s enormous data processing and transfer needs. But telecommunications providers will also integrate AI applications internally to streamline customer experiences and minimize manual intervention through self-service capabilities.

Since AI relies on massive amounts of high-quality data, telecom operators must collapse their internal data siloes to create a unified source of truth. Despite its transformative power, AI is only as good as the data you feed it.

Optical networking innovation is crucial in supporting AI’s massive data requirements. DWDM, which first increased fiber network bandwidth in the 1990s, remains vital as bandwidth demands increase due to the growth of internet traffic and AI applications.

However, service providers are increasingly integrating IPoDWDM (IP over DWDM) in long-haul and metro segments, allowing them to remove the transport (or DWDM) layer between routers to substantially reduce the long-term operational costs of IP-based networks.

Open optical networking will also prove critical for network evolution in the era of AI, allowing network operators to integrate 400G ZR and ZR+ coherent pluggable optics through open line systems to significantly increase capacity. This development is a game changer, particularly with Shannon’s limit (the maximum physical limit for how much traffic can pass through a fiber) already here.

Open line systems with an expanded L-Band option allow operators to double the capacity of a single fiber pair and overcome these physical limitations. We may also see further developments in higher baud rates, new modulation formats and even frequency bands outside of the C and L-Bands. While fiber's physical limitations remain, network requirements will heighten due to AI’s demands, so optical innovation is an essential part of the network’s continued evolution.

Forging Efficient and Sustainable Networks Through Automation

Sustainability is a relatively new goal in telecommunications. Historically, traffic throughput and CapEx costs were the industry’s focus, with sustainability being an afterthought, if it was even considered. In comparison, network automation is not novel, but its prohibitive costs and operational complexities were previously difficult to justify.

With the advent of AI, networks must be better, faster, and simpler. As a result, operators are leveraging advanced analytics to enhance network automation and efficiency, helping them improve sustainability by reducing power consumption. These tools have helped operators overcome automation’s previous limitations, empowering them with agile operations through real-time data insights.

Automation enables internet carriers to facilitate proactive network management through predictive analytics, ensuring the reliable low-latency connectivity enterprises need for real-time AI functionality.

Automation also enhances network scalability, keeping operators adaptable to AI applications’ resource requirements by automatically scaling capacity up or down according to real-time demands. Automation’s dynamic resource allocation and predictive maintenance capabilities also help operators mitigate AI’s massive energy consumption, enabling them to minimize energy waste resulting from low-demand periods or malfunctioning equipment.

Dream a Little Dream: Quantum Computing and Fixed Wireless Access

Telecommunications has always turned ambitious dreams into reality through technological innovation. Quantum computing could revolutionize the industry, potentially fitting the entire internet on a piece of equipment as small as a sugar cube.

While we likely won't see real-world integration for at least 10 years, quantum networking will progress as network demands rise and space and power become more precious. Quantum networking innovations may include enormous traffic capacities, specialized fiber types, further integration of quantum key distribution and other developments.

While quantum key distribution is already enhancing network security and efficiency, operators will realize quantum networking’s true potential when they can transport quantum traffic over long distances to serve global enterprises’ networking needs.

The digital divide has improved since the internet’s early days, but operators still have plenty of work left to enable equitable access to global connectivity. Fixed wireless access (FWA) will likely grow in the coming years, offering the most cost-effective connectivity option for underserved communities across the globe.

Equitable Internet access is a more attainable dream than quantum networking and other lofty technologies, but it requires substantial investment. Still, that investment is lower than the cost of fiber buildouts, making FWA an appealing option to bridge the digital divide and realize the dream of a fully interconnected world.

Putting the Pieces Together to Enable Evolution

When considering connectivity’s rapid progress over the past 30 years, network evolution is impossible to predict. While change is certain, the underlying drivers of network evolution will likely remain constant.

Whether enabling widespread AI integration or improving network automation, operators will continue to enhance high-capacity bandwidth, reliability, cost efficiency, scalability, reach and other familiar qualities.

These are the underlying pieces of the overall puzzle comprising connectivity's future. One thing is guaranteed, no matter how the network evolves over the next few decades, it will be a wild ride.

About the Author

Mattias Fridström | Vice President and Chief Evangelist of Arelion

Mattias Fridström is Vice President and Chief Evangelist of Arelion. With over 20 years in the telecommunications industry, Mattias can be considered a veteran. Since joining Telia in 1996, he has worked in a number of senior roles within Telia Carrier (now Arelion) and most recently as CTO. He has been Arelion's Chief Evangelist since July 2016. For more information, visit www.arelion.com. Follow Mattias on X @MFridstrom. Follow Arelion on X @ArelionCompany and LinkedIn.