As artificial intelligence (AI) continues to be integrated into everything, everywhere, 5G networks are starting to feel the strain—particularly when it comes to uplink traffic. A recent analyst forecast tipped uplink traffic to increase significantly in the coming years. And shifting traffic patterns mean mobile networks must adapt to avoid future bottlenecks.
“The main challenge with AI-driven applications is the sheer volume of data being generated and transmitted,” said Doug Wadkins, VP of product management and strategy at Opengear. Wadkins said Opengear is already seeing a shift toward more decentralized intelligence where processing happens closer to the edge of the network.
“Network optimization will become increasingly important, as machine learning algorithms dynamically adjust network configurations to enhance performance and manage resources efficiently,” he told Fierce Network. “The challenge is managing the ever-increasing complexity that comes with ensuring that our 5G infrastructure can keep pace with these rapid environments.”
Connected cars serve as a prime example of how AI applications are already influencing traffic patterns, noted Bob Everson, senior director of mobile architecture and ecosystem at Cisco.
While connected cars are presently the most common example of this, Everson said another affected use case is cameras and video surveillance. In some cases, automation is reducing the traffic sent to the cloud for smart cameras and moving processing to the edge. But while AI is helping to streamline data flows, this sort of optimization may only slow—not stop—the rising tide of uplink traffic as new workloads emerge.
“Beyond this, it remains to be seen what the broader impact will be,” Everson said.
The path to 6G
A report from Mobile Experts recently predicted that the rise of AI applications—particularly AI-powered assistants and augmented reality (AR) platforms—will soon drive mobile data traffic beyond the capacity of current 5G networks.
Joe Madden, principal analyst at Mobile Experts, warned that the mobile network could see capacity issues as early as 2027. The modern network is designed to handle much higher downlink traffic than uplink traffic.
“This arrangement works well for streaming video and web surfing, but new AI-Assistants such as Siri, Google Assistant and others will drive an increase in uplink traffic that is unprecedented,” the report noted.
With Mobile Experts’ predicted capacity shortfalls, the countdown to 6G has already begun.
“The network will be vulnerable to any ‘viral’ AR game or popular Gen-AI application in the 2028-2030 timeframe. By 2030, 6G capacity will be crucial to keep the network running,” Madden said.
Looking ahead to 6G, Wadkins said Opengear anticipates “a paradigm shift” with AI more deeply integrated into the network core.
“6G suggests that it will need to offer even greater bandwidth, lower latency and higher data speeds to accommodate future AI demands,” he said. Additionally, it may need to integrate AI-driven network optimization techniques to handle the enormous data influx expected from next-gen applications.
Preparing for the uplink surge
In the meantime, service providers and infrastructure companies must work to optimize network performance, manage uplink traffic and prepare for the growing influence of AI on 5G networks.
Increased network strain “drives increased complexity which increases management overhead and risk,” Wadkins said. He suggested providers can mitigate complexity with a neutral independent management network that can automate deployment and catalogs connected resource attributes and state. A catalog forms a source a truth for compliance, security, automation and remediation, while the out of band network is the neutral arbiter for IT infrastructure resources.
Cisco is also already addressing the need for service providers to manage an increasing flow of AI-generated data.
While much of AI processing currently happens on the device itself—particularly for privacy reasons—Everson noted that “cloud-hosted inference for AI models” will become more common, driving more demand for edge compute resources in the network.
“Given that we’re still in the early days of these technologies, we don’t yet know the full extent of the impact of AI applications impacting uplink traffic on 5G networks – though history shows that we’ll be able to optimize as we go,” Everson concluded. “Nonetheless, traffic patterns are shifting to more uplink than historical patterns.”
El artículo original puede consultarse en: