Four-stage hierarchical optimization to solve 5G network optimization challenges

With hundreds or even thousands of parameter combinations, it is almost impossible to find the optimal parameter value manually, and the difficulty of network optimization increases rapidly. On the other hand, in the case of tight wireless resources, how to allocate and schedule resources and how to effectively manage the relevant parameters of the wireless network are all challenges faced by network optimization.

The overall 5G network optimization method can be viewed from the following four perspectives. The first stage is initial optimization, the second stage is emergency network optimization, and the remaining two stages are continuous optimization and guided optimization. For any network optimization, the first step must be network engineering construction, followed by interference search. Interference search is extremely important. Once there is interference, the network signal performance will be limited. During the interference search, network quality comparison and network monitoring need to be performed simultaneously.

General checks during initial optimization phase

Generally, the final stage of deployment is the initial optimization stage, which mainly involves performance testing and performance tuning after deployment. After the base station is deployed, conventional antenna measurement, RF measurement and other functional tests are definitely necessary. In this stage, downlink/uplink data testing, delay testing, etc. are completed.

Considering the high complexity of 5G networks, the downlink direction of the network can be adjusted to a high data rate, which can also optimize the network configuration for low latency. At this stage, in order to complete the entire optimization process, the test vendor will perform network slicing so that some parts of the network prioritize high data rates and some parts prioritize low latency.

In this link, uplink interference is the first interference to be searched. As we all know, 5G is mainly deployed in TDD mode, so the uplink and downlink share the same frequency, but are staggered in time. Therefore, if the network is configured with a clear bias towards the downlink, there will be a large number of downlink time slots in the time domain (a small number of uplink time slots), and the uplink signal will hardly be seen in the spectrum analyzer at this time, making it difficult to see any interference. There are two important points in the optimization test of uplink interference, data throughput and RF parameters. By summarizing the test indicators of MCS and MIMO layers, this is the optimization work done by the base station, which can determine the final appropriate data packet length and MIMO layer, so as to optimize the data throughput in the uplink or downlink direction.

In TDD mode, the complexity of uplink interference optimization is that the downlink always covers the uplink. The SIB system information broadcast played by the base station can configure the uplink and downlink parameters, that is, the start and end points of the uplink time slot can be seen in the decoded SIB. With a little configuration, the uplink start and end information can be obtained, and then specific time gating can be provided. The downlink phase test and tuning are the same. Most of the related optimizations in the initial optimization phase are in performance testing and tuning.

What to do for emergency network optimization?

The second optimization phase, emergency network optimization, mainly conducts troubleshooting to solve problems in the network. The process is not complicated. Network management KPIs show network quality problems, and then emergency network optimization performs spectrum measurements, signal coverage verification, and other measurements.

If the display shows high BLER and low MCS, then there may be interference in the uplink direction. During the process, if the bit rate, delay and continuity can be integrated into a single result, the optimization will be more convenient and faster. This method sends a UDP packet stream from the terminal to the server, and then the server sends it back to the terminal, using the TWAMP protocol, which is a two-way active measurement protocol. The packet rate, packet size, and delay in this two-way test mode are all configurable and can be configured according to different network traffic requirements. Then the test results can integrate the bit rate, delay and continuity into the same result, optimizing the convenience of the entire process.

External interference sources often appear in emergency network optimization. Another potential interference source is TDD time synchronization, so time testing in synchronization is also essential.

Continuous optimization for efficient continuous network testing

Continuous optimization requires a continuous stream of data to perform this routine optimization, and cloud-based testing is more suitable for this type of continuous optimization, that is, the control and post-processing of the test equipment is transferred to the cloud, and these applications are accessed from any location connected to the Internet. Drive test equipment, benchmark test equipment, and indoor collection equipment can all be directly optimized using cloud-based testing. Cloud-based testing can bring more efficient data collection methods and can efficiently generate data without a lot of manual interaction.

The end-to-end data flow is also done by remotely configuring the entire test equipment, configuring the complete measurement file, and then the engineer pushes the task to the probe of the terminal to execute the task. The device regularly uploads high-level KPIs and some location and status updates to the management end. After the task test, these terminals will upload the file to a dedicated predefined file server, which is then processed by the automatic measurement file.

The most important thing to achieve this efficient and continuous network testing is undoubtedly reliable test equipment, especially the terminal probe, which must be maintained at a certain temperature to ensure its normal operation. This can be achieved with the help of passive cooling devices with forced convection, or active cooling devices with Peltier elements to keep the UE and probe at a constant temperature level.

Final Thoughts

The last part of guided optimization is mainly to deal with the increasingly complex market problems of 5G networks, analyze the large amount of data collected and provide users with guided optimization suggestions. Machine learning also plays a role in it, using different algorithms to further play the role of large amounts of data.

From the end of deployment to each stage in the network application link, the optimization of these four stages can quickly reduce the difficulty of 5G network optimization. When wireless resources are tight, resources can be reasonably allocated and scheduled, and relevant parameters of the wireless network can be effectively managed.