by Yatin Dharamshi, Head of Digital Operations – Orchestration and Fulfilment Engineering, Nokia
For many communications service providers (CSPs), self-organizing networks (SON) have been the golden key to efficiently configure and optimize booming mobile networks with closed-loop automation. SON has brought tremendous value for earlier-generation mobile technologies. And if they were not already essential back then, they are undoubtedly going to be crucial for CSPs to manage the complexity that comes with the adoption of 5G technology.
So why does 5G make self-organizing networks more exciting in the next decade compared to the last? To start, 5G is a wireless technology that promises to cut the cord – untethering people and objects from certain locations or places. To achieve this promise, 5G has introduced a slew of new capabilities and deployment options to wireless networks. These emerging technologies include network slicing, dynamic spectrum sharing, beamforming, edge cloud, orchestration and more.
On its own, these technologies already pose difficulties for humans to manage individually. And if these technologies were united, the challenges that arise are becoming more difficult for humans to control and manage with the current toolset at our disposal. Further, 5G offers multiple frequency bands, which combined with network slicing would unleash a myriad of use cases. But with growing number of use cases (and technologies), complexity increases. This is where closed-loop automation comes in, and where SON will shine. Now more than ever, self-organizing networks will be a critical element in the shift toward autonomous operations, which will push current-generation SON to its limits.
A snapshot of SON and its capabilities
Most common self-organizing networks today have five key abilities. First, it drives automation and reduces the reliance on manual operational applications starting from network configurations during rollout to keeping networks optimized thereafter.
Second, SON can conduct rapid, real-time detection of cell outages or degraded performance. This is crucial for network operators and CSPs to ensure that their networks can efficiently cope with unprecedented loads, as well as calibrate nearby cells to balance out the lost coverage.
The third capability of SON is that it allows for seamless connectivity, which helps network operators achieve optimal performances and overcome challenges around insufficient capacity or coverage and mobility robustness.
Fourth, self-organizing networks can carry out ongoing network monitoring and healing. A typical example is identification of “sleeping cells,” or cells performing sub-optimally, and instinctively reset it to improve network reliability.
Lastly, SON is great for cost management as it enables network operators to manage and control costs. For e.g., by optimizing a network’s energy consumption. This is done through the active monitoring of cell loads so that traffic cells can be switched on or off automatically when needed.
Upgrading with cognitive SON
Today, many higher-order self-organizing network functions require a human expert touch. By this, I mean the involvement by experienced — and at times, hard to come by — optimization engineers in the following tasks:
- Identify and place network performance objectives
- Evaluate network conditions across individual regions and parameters, such as rural vs. urban, high-volume vs. low-volume, and so on
- Analyze and correct problems, while also determining if those corrections were effective
While having the eyes of a human expert on SON functions is great, the dependency can also create bottlenecks in the dynamic and radically complex 5G environment. Besides, as humans, we are naturally prone to errors. Thus, shifting from human-led automation to fully machine-led autonomous operations is key.
This is where the next generation of self-organizing networks with cognitive abilities come into play. Cognitive SON is ideal because it brings in machine learning to take over manually driven SON functions. To do this, a mobile operator simply sets the objectives, and cognitive SON will do the rest: understanding network context, identifying problems, applying and orchestrating the right actions, and evaluating their efficacy. Machine learning is truly the secret ingredient to cognitive SON’s effectiveness. Its intelligence allows for predictive analytics, so cognitive SON can characterize networks, label different cells based on the deployment area and problems present, and instinctively invoke proper algorithms that provide solutions to reach an objective – all without the need for human intervention.
To take your cognitive SON to the next level, moving certain functions to the edge cloud will be key as it reduces latencies. Mobile operators understand that common networks today require a long time to collect data, and it’s often not real-time or near-real-time. But on the edge cloud, real-time data collection becomes a reality, which allows for faster changes in controls or functions, thus achieving swifter reaction times to problems that may arise. Combine this capability with the predictive analytics brought forth by machine learning, and cognitive SON becomes an extremely powerful tool.
Take a leap of faith in cognitive SON
The benefits of cognitive self-organizing networks are abundant. Its intelligent automation capabilities deliver improved, more consistent customer experiences, while also ensuring timely and automatic problem detection so issues can be mitigated in digital-time.
It’s understandable that not every CSP will be ready or comfortable to leap from their current human-led network optimization operations to trusting a fully machine-led autonomous system overnight. So, to ease CSPs into this new mode of working, cognitive SON offers extensive visibility and open controls so that experts can have a strong hand in influencing its operation’s journey while gradually adopting a fully autonomous system.
As we move deeper into the 5G era and CSPs continue on their journey to digitization, cognitive SON will be the key step for enabling autonomous intelligent closed-loop systems, and therefore achieving successful 5G operations.
