Picked from the software, topic, and level of “Class 5G Resource Allocation Deep RL #1: Concept of 5G/6G Resource Allocation: Resource Block, Power, User”.
At most one email a week. Unfollow anytime with one click.
Transcript
The full narration of this video. The current sentence is highlighted while the video plays; click any sentence to jump there.
Show transcript
Introduction
This is the final result, basic network 5G calculation with Python. We see how SNR and throughput of one resource block decrease with distance, and find an interesting fact, users close to the tower can be served with much smaller power without losing speed. Let's start. Hello, welcome to the research tutorial series, optimizing 5G and 6G network resource allocation with Deep Reinforcement Learning. Every half millisecond, a 5G base station must decide which users get a resource block and how much power they get. In this 6-episode series, we build everything from scratch. Resource concepts, network simulator, classic scheduler, MDP formulation, DeepQ Network agent, up to throughput evaluation, fairness, and energy efficiency. In the first episode, we understand the resources. We open the terminal through the terminal menu, new terminal, then install nampai, tork, gymnasium, and matplotlib.
Step 1: Installing the library
PyTorch for the QN neural network, and gymnasium for the reinforcement learning environment.
Step 2: Numerology, RB & Power
Create a parameter .py file We use the N78 band at 3.5 GHz with numerology MU equal to 1, subscriber spacing of 30 kHz. One resource block consists of 12 subcarriers, so its bandwidth is 360 kHz, and one slot lasts half a millisecond. We schedule 10 resource blocks. Base station power is 40 dBm or 10 Watts, so 1 Watt per resource block, with noise figure 9 dB and maximum spectral efficiency of 7.4 bits per second per hertz for 256 QAM.
Step 3: Noise, path loss & Shannon
The draw RB function calculates thermal draw. Minus 174 dBm per hertz plus 10 log bandwidth plus noise figure. The path loss function uses the 3GPP TR 38901 urban macro model without line of sight, and the RB rate function calculates throughput from the Xenon capacity, limited by the spectral efficiency of 256 QAM. Create a test underscore concept.py file and run it.
Step 4: SNR & throughput versus distance
10 resource blocks wide 3.6 MHz, with draw minus 109.4 dBm per RB. At 50 meters, SNR is 48.6 dB, at 150 meters 30 dB, both have reached the limit of 2.66 Mbit per second per RB. At 300 meters it drops to 2.19, and at 5,000 meters 1.2 Mbit per second. Interestingly, at 300 meters, cutting power in half only reduces throughput by 16%. Let's look at the graph.
Step 5: 256-0AM limits & resource grid
SNR drops sharply with distance. On the throughput graph, the flat curve at the beginning is the limit of 256 kbit. Users close by are already saturated, so even 50 or 25% power results in the same speed. This is the energy-efficient opportunity that the QN agent will later study. The right graph illustrates the resource grid. Each box is one resource block in one slot, and the color represents the user being served.
Done!
Done. We have now understood resource blocks, slots, transmit power, draw, path loss, and CNN capacity in 5GNR. In the next episode, we will build a complete cellular network simulator with fading, interference, and SINR. Currently working on a thesis, dissertation, or research in the field of telecommunications and artificial intelligence? Cerdas Inovasi Digital helps implement cellular network research, from 5G simulators, channel models and SINR, proportional fair and waterfilling scheduling, up to pre-inforcement learning with PyTorch, as well as opening Python programming courses for electrical engineering and telecommunications. Want to master the entire series, from resource block concepts to evaluating the energy efficiency of the QN agent? All 6 episodes, complete with PDF step-by-step guides and source code, are available to premium tutorial subscribers. Subscribe starting at Rp69,000 per month, save 28% with the 12-month package, or pay once Rp999,000 for lifetime access. Want just one class? Buy the single class with lifetime access. Subscribe now at the premium page cerdasinovasidigital.com. Contact us via WhatsApp, or visit cerdasinovasidigital.com. Thank you for watching, see you in the next episode.
Ask the class material
Type a question or keyword; we find the part of this episode that explains it and jump the video there.
My notes
Notes are saved with the current video time and are only visible to you.
Questions & answers
Stuck on a step? Ask here; the instructor and other learners can answer.
No questions yet.
Your instructor
Cerdas Inovasi Digital Team
Software developers, BIM & 3D engineers, and AI practitioners
53 video classes1,290 episodes
We build software, BIM models, games and AI automations for clients every day, and turn that hands-on work into step-by-step classes: every step is shown on screen and comes with a PDF guide you can keep.
