Scribd
Upload
18 views36 pages

Network Slicing

The document discusses network slicing as a method to allocate network resources based on the varying demands of different devices, such as high bandwidth for mobile phones and low latency for self-driving cars. It highlights the challenges of resource allocation due to complexity, privacy laws, and limited resources, and proposes the use of artificial intelligence and machine learning to dynamically adapt to changing user requirements. Additionally, it introduces deep reinforcement learning as a potential solution for optimizing network slicing.

Uploaded by

Maxwell Aniakor
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
18 views36 pages

Network Slicing

The document discusses network slicing as a method to allocate network resources based on the varying demands of different devices, such as high bandwidth for mobile phones and low latency for self-driving cars. It highlights the challenges of resource allocation due to complexity, privacy laws, and limited resources, and proposes the use of artificial intelligence and machine learning to dynamically adapt to changing user requirements. Additionally, it introduces deep reinforcement learning as a potential solution for optimizing network slicing.

Uploaded by

Maxwell Aniakor
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 36

Network slicing for future

fixed and 6G mobile


broadband Architecture
Maxwell Aniakor
B.Eng Electronics and Communication Engineering, UK.
Data Professional [MIT Data Science and Statistics]
12 Yrs. Telco World
Nokia, Aviat, Huawei
Phone: 07085613967
Email: [email protected]
Global mobile traffic in 2020-2030 forecast by ITU
Different dimensions of the architecture of green 6G
What is Network Slicing
Self-driving cars or equipment used for remote surgery will need
fast transmission of information. These devices will need a very
low delay time or latency.
What is Network Slicing

Things like mobile phones, VR headsets or laptops might need a high bandwidth,
which means more data being sent or received per second.
What is Network Slicing

In contrast, low energy devices like sensors, smart fridges or other Internet of
Things devices have low data requirements, but they need many connections,
maybe millions per square kilometre.
What is Network Slicing

Now, there are many different devices


out there, but they tend to fit into
different categories depending on their
demands.

So it makes sense that we virtually


separate the different user types
depending on what resources they
need and this is called network slicing.

It's basically a way for the network


providers to make sure that each user
type has the right quality of service for
their needs.
eMBB: Enhanced Mobile Broadband
URLLC: Ultra-Reliable and Low Latency
mMTC: Massive Machine-Type Communications
1st Challenge
 Although network slicing makes things easier it's still pretty challenging to share out resources to
different slices.
 This is because the network will be really complex and it'll be changing constantly.
 `Just think of how many users in each slice will be demanding different resources at different times.
 How do you process all those requests as quickly as possible?
2nd Challenge
 This is made even more challenging by privacy laws.
 Network providers aren't allowed access to your user requirements, so it's difficult to get
accurate information about what resources are required and when.
3rd Challenge

 In addition, all these network resources are limited.

 You just can't give every device a super high bandwidth and low latency service.

 It would cost too much energy and too much money.


=
We are going to see something a lot faster and a lot Smatter.
Here Comes …

Artificial Intelligence Powered Network


Slicing
Why use Machine Learning to
solve network slicing problem ?
The main reason is because the network is really
dynamic. It's constantly changing due to the changing
requirements of many different users and applications.

Traditional algorithms are designed for a specific


scenario or application and do not tend to work well
for constantly changing environments.

The network system is constantly producing lots of


useful data that shows all these changes in real time.

So we really need to use something that can learn from


this data and adapt with any changes. This is why
machine learning is so useful.
What is Machine Learning ?
Machine learning is a tool to learn
knowledge from the data directly.

You just show it data and tell it what you are


going to achieve with that data.

It can also use that data to make predictions to


some extent.

And machine learning is able to detect a


change in the data in real time or near real
time, and can apply the knowledge learned
from this change to allocate resources quickly.
Hence they meet the user application and the network requirements.

It's a good way for the users who get the network resources they need.

It's also good for the network operators. So if the service they provide is more reliable, then they are
fulfilling the requirements of their service level agreement and then they don't suffer a financial penalty.
What is Deep Learning ?

The official definition is: Deep Learning (DL) is a machine learning


method using artificial neural networks.
What is Deep Learning ?
What is Deep Learning ?
We can’t Just rely on Deep Learning
Reinforcement Learning
Reinforcement Learning

https://arxiv.org/ftp/arxiv/papers/1912/1912.05744.pdf
DEEP REINFORCEMENT LEARNING
DEEP REINFORCEMENT LEARNING = DEEP
LEARNING + REINFORCEMENT LEARNING
AlphaGo’s DRL beat’s world’s best Go Player
Thank You

You might also like