چكيده به لاتين
In 4th and 5th generation networks, network activities will be implemented in software and cloud. The work done has used existing platforms to evaluate protocol, architecture, virtualization platform. However, improving software code will also lead to the improvement of the results in the above works.
Conventional optimization methods in software systems, such as use of debuggers, logging, and monitoring systems, will not be suitable for a real-time application specially in production environment, as they will create a lot of overhead on the system and software.
In this study, a method based on system observability was presented. This method used to identify system bottlenecks and extract its information in a production environment. This information can be feed to management systems for decision making such as orchestration, resource forecasting algorithms. To improve the performance of the software radio area, the proposed method was chosen to be compatible with the real-time nature of radio access network and have the least overhead and negative impact on such operations. Profiling and Tracing were used for this purpose.
In order to evaluate the usefulness of this method in comparison with other conventional methods, the existing results were compared with the results obtained from this method. Processing time of possible bottlenecks activities were measures. The results showed that the proposed method with special tools can achieve similar results without the need of more resource compared to other conventional methods.
The results obtained in this study showed that in the protocol stack of the radio network, the activities of the physical layer spend the most processing time. These processes are divided into two categories: cell processes and user process. The results of other work have shown that cell processing takes the most time. And for such processes, our work showed that the activities of the coding and modulation unit need the most time to complete and have higher complexity compare to other functions. We Also finds out that Uplink processes have higher CPU time compare to Downlink processes. It was also found that the effect of the hardware resources on the processing time of Layer 1 process are much higher that other layers in Radio Access Network protocol stack.
The proposed steps based on observability brings advantages such as no need of consuming too many resources. Thus, can be very useful to be use in containers. Also, this method does not interfere with the real-time operation of operations. Which makes the data more realistic.
