Showing posts with label Artificial Intelligence. Show all posts
Showing posts with label Artificial Intelligence. Show all posts

Wednesday, 6 February 2019

AI in 5G – the why and how

IET recently held the 6th Annual 5G conference bringing together key players in the 5G world. You can watch the videos for that event here (not all have been uploaded at the time of writing this post).

We reached out to Dr. Yue Wang to share her presentation with us and she has kindly done so. The presentation and video are embedded below.






Related posts:

Wednesday, 23 January 2019

AI and Analytics Based Network Designing & Planning

Recently I blogged about how Deutsche Telekom is using AI for variety of things. The most interesting being (from this blog point of view), fiber-optic roll-out. According to their press release (shortened for easy reading):

"The shortest route to the customer is not always the most economical. By using artificial intelligence in the planning phase we can speed up our fiber-optic roll-out. This enables us to offer our customers broadband lines faster and, above all, more efficiently," says Walter Goldenits, head of Technology at Telekom Deutschland. It is often more economical to lay a few extra feet of cable. That is what the new software-based technology evaluates using digitally-collected environmental data. Where would cobblestones have to be dug up and laid again? Where is there a risk of damaging tree roots?

The effort and thus costs involved in laying cable depend on the existing structure. First, civil engineers open the ground and lay the conduits and fiber-optic cables. Then they have to restore the surface to its previous condition. Of course, the process takes longer with large paving stones than with dirt roads.

"Such huge amounts of data are both a blessing and a curse," says Prof. Dr. Alexander Reiterer, who heads the project at the Fraunhofer IPM. "We need as many details as possible. At the same time, the whole endeavour is only efficient if you can avoid laboriously combing through the data to find the information you need. For the planning process to be efficient the evaluation of these enormous amounts of data must be automated." Fraunhofer IPM has developed software that automatically recognizes, localizes and classifies relevant objects in the measurement data.

The neural network used for this recognizes a total of approximately 30 different categories through deep learning algorithms. This includes trees, street lights, asphalt and cobblestones. Right down to the smallest detail: Do the pavements feature large pavement slabs or small cobblestones? Are the trees deciduous or coniferous? The trees' root structure also has a decisive impact on civil engineering decisions.

Once the data has been collected, a specially-trained artificial intelligence is used to make all vehicles and individuals unidentifiable. The automated preparation phase then follows in a number of stages. The existing infrastructure is assessed to determine the optimal route. A Deutsche Telekom planner then double-checks and approves it.


In the recent TIP Summit 2018, Facebook talked about ‘Building Better Networks with Analytics’ and showed off their analytics platform. Vincent Gonguet, Product Manager, Connectivity Analytics, Facebook talked about how Facebook is using a three-pronged approach of accelerating fiber deployment, expanding 4G coverage and planning 5G networks. The video from the summit as follows:

TIP Summit 2018 Day 1 Presentation - Building Better Networks with Analytics from Telecom Infra Project on Vimeo.

Some of the points highlighted in the video:
  • Educating people to connect requires three main focus areas, Access, Affordability and Awareness – One of the main focus areas of TIP is access. 
  • 4G coverage went from 20% to 80% of world population in the last 5 years. The coverage growth is plateauing because the last 20% is becoming more and more uneconomical to connect.
  • Demand is outpacing supply is many parts of the world (indicating that networks has to be designed for capacity, not just coverage)
  • 19% of 4G traffic can’t support high quality videos today at about 1.5 Mbps
  • Facebook has a nice aggregated map of percentage of Facebook traffic across the world that is experiencing very low speeds, less than 0.5 Mbps
  • Talk looks at three approaches in which Facebook works with TIP members to accelerate fiber deployment, expand 4G coverage and plan 5G networks.
  • A joint fiber deployment project with Airtel and BCS in Uganda was announced at MWC 2018
  • 700 km of fiber deployment was planned to serve over 3 million people (Uganda’s population is roughly 43 million)
  • The real challenge was not just collecting data about roads, infrastructure, etc. New cities would emerge over the period of months with tens of thousands of people 
  • In such situations it would be difficult for human planners to go through all the roads and select the most economical route. Also, different human planners do thing in different ways and hence there is no consistency. In addition, its very hard to iterate. 
  • To make deployments simpler and easier, it was decided to first provide coverage to people who need less km of fiber. The savings from finding optimal path for these people can go in connecting more people.
  • It is also important for the fiber networks to have redundancy but it’s difficult to do this at scale
  • An example and simulation of how fiber networks are created is available in the video  from 07:45 – 11:00.
  • Another example is that of prioritizing 4G deployments based on user experience, current network availability and presence of 4G capable devices in partnership with XL Axiata is available in the video from 11:00 – 14:13. Over 1000 sites were deployed and more than 2 million people experienced significant improvement in their speeds and the quality of videos. 
  • The final example is planning of 5G mmWave networks. This was done in partnership with Deutsche Telekom, trying to bring high speeds to 25,000 apartment homes in a sq. km in the center of Berlin. The goal was to achieve over 1Gbps connection using a mixture of fiber and wireless. The video looks at the simulation of Lidar data where the wireless infrastructure can be deployed. Relevant part is from 14:13 – 20:25.
Finally, you may remember my blog post on Automated 4G / 5G Hetnet Design by Keima. Some of the work they do overlaps with both examples above. I reached out to Iris Barcia to see if they have any comments on the two different approaches above. Below is her response:

“It is very encouraging that DT and Facebook are seeing the benefits of data and automation for design. I think that is the only way we’re going to be able to plan modern communication networks. We approach it from the RAN planning perspective: 8 years ago our clients could already reduce cost by automatically selecting locations with good RF performance and close to fibre nodes, alternatively locations close to existing fibre routes or from particular providers. Now the range of variables that we are capable of computing is vast and it includes aspects such as accessibility rules, available spectrum, regulations, etc. This could be easily extended to account for capability/cost of deploying fibre per type of road. 

But also, we believe in the benefit of a holistic business strategy, and over the years our algorithms have evolved to prioritise cost and consumers more precisely. For example, based on the deployment needs we can identify areas where it would be beneficial to deploy fibre: the study presented at CWTEC showed a 5G Fixed Wireless analysis per address, allowing fibre deployments to be prioritised for those addresses characterised by poor RF connectivity.”

There is no doubt in my mind that more and more of these kinds of tools that relies on Analytics and Artificial Intelligence (AI) will be required to design and plan the networks. By this I don’t just mean 5G and other future networks but also the existing 2G, 3G & 4G networks and Hetnets. We will have to wait and see what’s next.


Related Blog Posts:

Friday, 26 October 2018

The Yin and the Yang of AI & Blockchain


Today I read about HTC's Exodus 1, new Blockchain smartphone that only people with crypto-currency can buy. SCMP described in very simple terms what this phone is for:

Both HTC’s Exodus and Sirin’s Finney smartphones feature a built-in digital wallet application that will enable users to securely store and use cryptocurrencies, such as bitcoin and ethereum, in daily transactions.

Those smartphones are designed to replace the special memory sticks, which employ complex usernames and passwords to access, that cryptocurrency investors use to store their digital money. These investors typically store most of their cryptocurrencies in such hardware, which are kept offline as a means of security.

“There are things that a phone manufacturer can do with a chip that nobody else can,” said Chen. “We want to be safer than the existing hardware wallets … HTC has a track record of making trusted hardware.”

The company’s Exodus smartphone, for example, can serve as a “node”, which can connect to certain blockchain networks to enable trading of tokens between users. It will also be able to act as a so-called mining rig for users to earn new tokens tied to the Exodus blockchain.

“At some point, we’ll do our own utility token,” said Chen, adding that there was no timetable for such a token release.

HTC’s foray into blockchain, the distributed ledger technology behind cryptocurrencies like bitcoin, represents a strategy to keep the company relevant in smartphones, which is a market dominated by Samsung Electronics and Apple, followed by Huawei Technologies, Xiaomi and other major Chinese brands.

Anyway, the blockchain smartphone reminds me of the joke above (via marketoonist). The second technology mentioned in this joke is AI or Artificial Intelligence.

I heard HP Enterprise talk about AI recently and this picture above is a nice simple way to show how Deep Learning (DL), Artificial Neural Networks (ANN), Machine Learning (ML) and Artificial Intelligence (AI) are related.

I see AI and blockchain often referred to together. This does not necessarily mean that they are related. iDate allowed me to share a recent presentation (embedded below) that refers to AI & blockchain as Yin and Yang. Anyway, I am happy to learn more so if you have any thoughts please feel free to share.



Further Reading:


Related Posts:

Tuesday, 2 October 2018

Benefits and Challenges of Applying Device-Level AI to 5G networks


I was part of Cambridge Wireless CWTEC 2018 organising committee where our event 'The inevitable automation of Next Generation Networks' covered variety of topics with AI, 5G, devices, network planning, etc. The presentations are available freely for a limited period here.

One of the thought provoking presentations was by Yue Wang from Samsung R&D. The presentation is embedded below and can be downloaded from Slideshare.



This presentation also brought out some interesting thoughts and discussions:

  • While the device-level AI and network-level AI would generally work cooperatively, there is a risk that some vendor may play the system to make their devices perform better than the competitors. Something similar to the signaling storm generated by SCRI (see here).
  • If the device-level and network-level AI works constructively, an operator may be able to claim that their network can provide a better battery life for a device. For example iPhone XYZ has 25% better battery life on our network rather than competitors network.
  • If the device-level and network-level AI works destructively for any reason then the network can become unstable and the other users may experience issues. 

I guess all these enhancements will start slowly and there will be lots of learning in the first few years before we have a stable, mutually beneficial solution.

Related Posts:

Monday, 13 August 2018

Telefonica: Big Data, Machine Learning (ML) and Artificial Intelligence (AI) to Connect the Unconnected


Earlier, I wrote a detailed post on how Telefonica was on a mission to connect 100 Million Unconnected with their 'Internet para todos' initiative. This video below is a good advert of what Telefinica is trying to achieve in Latin America


I recently came across a LinkedIn post on how Telef√≥nica uses AI / ML to connect the unconnected by Patrick Lopez, VP Networks Innovation @ Telefonica. It was no brainer that this needs to be shared.



In his post, Patrick mentions the following:

To deliver internet in these environments in a sustainable manner, it is necessary to increase efficiency through systematic cost reduction, investment optimization and targeted deployments.

Systematic optimization necessitates continuous measurement of the financial, operational, technological and organizational data sets.

1. Finding the unconnected


The first challenge the team had to tackle was to understand how many unconnected there are and where. The data set was scarce and incomplete, census was old and population had much mobility. In this case, the team used high definition satellite imagery at the scale of the country and used neural network models, coupled with census data as training. Implementing visual machine learning algorithms, the model literally counted each house and each settlement at the scale of the country. The model was then enriched with crossed reference coverage data from regulatory source, as well as Telefonica proprietary data set consisting of geolocalized data sessions and deployment maps. The result is a model with a visual representation, providing a map of the population dispersion, with superimposed coverage polygons, allowing to count and localize the unconnected populations with good accuracy (95% of the population with less than 3% false positive and less than 240 meters deviation in the location of antennas).


2. Optimizing transport



Transport networks are the most expensive part of deploying connectivity to remote areas. Optimizing transport route has a huge impact on the sustainability of a network. This is why the team selected this task as the next challenge to tackle.

The team started with adding road and infrastructure data to the model form public sources, and used graph generation to cluster population settlements. Graph analysis (shortest path, Steiner tree) yielded population density-optimized transport routes.


3. AI to optimize network operations


To connect very remote zones, optimizing operations and minimizing maintenance and upgrade is key to a sustainable operational model. This line of work is probably the most ambitious for the team. When it can take 3 hours by plane and 4 days by boat to reach some locations, being able to make sure you can detect, or better, predict if / when you need to perform maintenance on your infrastructure. Equally important is how your devise your routes so that you are as efficient as possible. In this case, the team built a neural network trained with historical failure analysis and fed with network metrics to provide a model capable of supervising the network health in an automated manner, with prediction of possible failure and optimized maintenance route.

I think that the type of data driven approach to complex problem solving demonstrated in this project is the key to network operators' sustainability in the future. It is not only a rural problem, it is necessary to increase efficiency and optimize deployment and operations to keep decreasing the costs.


Finally, its worth mentioning again that I am helping CW (Cambridge Wireless) organise their annual CW TEC conference on the topic 'The inevitable automation of Next Generation Networks'. There are some good speakers and we will have similar topics covered from different angles, using some other interesting approaches. The fees are very reasonable so please join if you can.

Related posts:

Tuesday, 13 February 2018

Artificial Intelligence - Beyond SON for Autonomous Networks


What is the next step in evolution of SON? Artificial Intelligence obviously. The use of artificial intelligence (AI) techniques in the network supervisory system could help solve some of the problems of future network deployment and operation. ETSI has therefore set up a new 'Industry Specification Group' on 'Experiential Networked Intelligence' (ISG ENI) to develop standards for a Network Supervisory assistant system.


The ISG ENI focuses on improving the operator experience, adding closed-loop artificial intelligence mechanisms based on context-aware, metadata-driven policies to more quickly recognize and incorporate new and changed knowledge, and hence, make actionable decisions. ENI will specify a set of use cases, and the generic technology independent architecture, for a network supervisory assistant system based on the ‘observe-orient-decide-act’ control loop model. This model can assist decision-making systems, such as network control and management systems, to adjust services and resources offered based on changes in user needs, environmental conditions and business goals.


The introduction of technologies such as Software-Defined Networking (SDN), Network Functions Virtualisation (NFV) and network slicing means that networks are becoming more flexible and powerful. These technologies transfer much of the complexity in a network from hardware to software, from the network itself to its management and operation. ENI will make the deployment of SDN and NFV more intelligent and efficient and will assist the management and orchestration of the network.


We expect to complete the first phase of ENI work in 2019. It will include a description of use cases and requirements and terminology, including a definition of features, capabilities and policies, which we will publish in a series of informative best practice documents (Group Reports (GRs)).
This will of course require co-operation from many different industry bodies including GSMA, ITU-T, MEF, IETF, etc.

Will see how this goes.

Further reading: