The evolution of network technology has been driven by a quest for unparalleled connectivity, outstanding speed, and exceptional data processing efficiency. With the rise of 5G, one of the most highly anticipated concepts was edge computing, which aimed to significantly reduce latency and offer premium services. However, despite the initial excitement and optimism, the broad deployment of edge computing by mobile network operators (MNOs) has remained largely underwhelming.
The Promise of Edge Computing
Aspirations and Expectations
Edge computing, also known as Mobile Edge Computing (MEC), emerged with high hopes that mobile network operators could position computational resources much closer to their base stations. This proximity was believed to enable ultra-low latency, providing the opportunity to charge higher prices for premium services. The anticipated benefits of such advancements included real-time data processing, enhanced consumer experiences, and the potential for innovative applications that demand fast response times.
Despite these expectations, the practical implementation of edge computing has not lived up to the hype. While relocating computation closer to users theoretically should cut latency, in reality, the benefits achieved have been marginal. Reducing response time by a millisecond or two does not significantly contribute to noticeable improvements. With fiber optic cables transmitting data at about 200 kilometers per millisecond and typical 5G networks having latencies ranging from 30 to 40 milliseconds, the impact of edge computing’s latency reduction remains limited.
Real World Implementation
In practice, the conventional model where telecommunications networks transport data to internet exchange points remains prevalent. These points are critical for facilitating the connection and data exchange between various internet service providers and networks. From these peering points, data is routed to large hyperscalers’ data centers for processing before responses make their way back to the originating devices. This existing model boasts better economies of scale due to the substantial size of hyperscaler-owned data centers and their superior ability to sell computing services over MNOs.
Despite the occasional trials conducted by MNOs, such as limited partnerships with search engine companies, the broad adoption of edge computing has been sparse. Most ongoing edge solutions resemble the traditional model, where traffic is still sent to centralized hyperscaler data centers for processing. MNOs have found that this centralized approach continues to present the most economically viable path for delivering services.
Current State and Challenges
Limited Trials and Adoption
The trials and adoption of edge computing by MNOs have been limited. While there have been instances where operators worked with computing giants to manage data locally, these cases are few and far between. Instead of using their computational resources, MNOs often default to established methods where data is transmitted to hyperscalers’ data centers for processing. Notably, in private 5G networks, edge computing instances are more of a technical formality than a broad adoption of the strategy, with traffic managed within a local IT network instead.
These limitations are compounded by the economic realities and practical challenges that MNOs face. Investing heavily in the deployment of edge computing infrastructure without seeing immediate and significant revenue opportunities proves to be a daunting risk. Centralizing the baseband processing for multiple base stations into single, more extensive, centralized units allows MNOs to reap the benefits of economies of scale and operational efficiency, rather than dispersing these resources near the network edge.
Economic and Operational Realities
The economic aspects illustrate a significant hurdle in edge computing deployment. Revenue generation from edge computing has not proven compelling for MNOs. The high costs associated with building, maintaining, and operating local edge data centers do not justify marginal latency improvements. Therefore, MNOs find centralizing their computational needs far more efficient and cost-effective than spreading the resources closer to the network edge.
Additionally, the technical and operational challenges of managing distributed computational resources introduce complexities that further deter MNOs from pursuing widespread edge computing. In contrast, centralized data centers already possess the infrastructure, experience, and scale necessary to manage large volumes of data processing tasks more effectively. This centralized approach aligns better with the existing operational models and revenue structures that MNOs are accustomed to.
Prospects for 6G Networks
The Undefined Future of 6G
Looking ahead to the advent of 6G technology, questions surrounding the role and viability of edge computing persist. While there are speculations about 6G potentially bringing “hyper 5G” capabilities—such as enhanced speeds and integrating AI-native functionalities—the path to implementing such technology on a broad scale remains unclear. The visions include incorporating sensing and AI capabilities, but these advancements need to be weighed against realistic deployment challenges and economic feasibility.
For edge computing to become mainstream within a 6G framework, several significant criteria must be met. There must be the emergence of new ultra-low-latency applications that demand latencies below 5 milliseconds, coupled with a market willingness to pay a premium for these advanced services. Furthermore, sufficient spectrum allocation to support low-latency air interfaces, coupled with extensive geographical deployment of 6G, would be essential.
Criteria for Mainstream Adoption
As it stands, meeting these criteria seems improbable. Many potential 6G applications echo promises initially associated with 5G that have yet to materialize. The willingness to pay premiums for 5G services remains minimal, posing challenges to establishing a commercial base for edge computing in a 6G scenario. Securing additional spectrum for 6G is another hurdle, as regulatory and technical constraints become increasingly difficult to navigate.
Presently, mid-band 5G (3.5 GHz) has only been deployed across about 20% of geographical areas in most countries. This limited coverage casts doubt on the feasibility of achieving extensive 6G deployment, which would be necessary for making edge computing commercially viable. Until these fundamental criteria are addressed, the broad adoption of edge computing will remain constrained by economic and operational realities.
Practical Applications and Market Demand
Sensing and AI Applications
Among the applications discussed for 6G, sensing and AI are noteworthy. However, the market demand for sensing applications remains speculative and unclear. Implementing sensing in a 6G context may require high-frequency spectrum, which is typically poorly suited for traditional communication purposes due to its limited range and penetration capabilities. The practical feasibility and commercial viability of such applications are yet to be demonstrated convincingly.
Regarding AI, while these applications often demand rapid response times, they generally either run directly on mobile devices (on-device processing) or require high-performance computational capabilities best provided by large data centers. Consequently, there is scant demand for the marginal 1-millisecond speedup that edge computing offers when using MNO networks. The established processing power and efficiency of hyperscaler data centers continue to overshadow the benefits of marginal latency improvements offered by edge computing.
Hyperscaler Dominance
The evolution of network technology has long been fueled by the pursuit of unparalleled connectivity, exceptional speed, and superior data processing efficiency. When 5G technology started gaining momentum, one of the most eagerly awaited advancements was edge computing. This innovation promised to drastically cut down latency and offer premium services by bringing computing resources closer to end-users. However, despite the initial enthusiasm and high expectations, the widespread adoption of edge computing by mobile network operators (MNOs) has been somewhat disappointing. The expansive deployment that many had anticipated has not yet materialized, and the reality has not matched the early optimism. While edge computing still holds considerable promise, its integration into the existing network infrastructure has faced several obstacles, leading to slower-than-expected rollout. The anticipated revolution in reduced latency and improved service quality remains more of a work in progress, rather than a completed leap forward in network technology.