This project is focused on optimization of multilayer application-aware networks. The key goal of the project is to develop, implement, and analyze models and algorithms for optimization of multilayer application-aware networks. An application-aware network can be defined as a network that is able to identify and classify applications and then use suitable optimization techniques to provision these application using resources accessible in the network in order to achieve acceptable application performance metrics. In turn, a multilayer network is a network modeled as a set of separate layers using various technologies and protocols applied to transmit data. In the context of this project, we assume that the network consists of two layers: packet layer and optical layer. The packet layer is used to directly serve the applications, i.e., to establish in the network demands required to serve various types of applications. In turn, the optical layer is used to establish virtual topologies to provision flows aggregated over the packet layer service demands.
The network optimization can be performed considering each single layer of the network separately. However, multilayer optimization allows to improve the overall network performance, namely, reduce CAPEX expenditures and at the same time holding the QoS (Quality of Service) parameters required by the applications by identifying the network configurations that jointly optimizes packet and optical resources. The idea of application-aware networking has been known for many years, however the vast majority of previous works in this area have been focused on a single layer (packet layer) optimization, i.e., the routing algorithms have been aware only of the packet layer resources and constraints. While at the same time, the configuration of optical layer resources have not been updated according to changing requirements. Moreover, recent trends in application characteristics, namely, the increasing diversity of requirements in almost all performance indicators (e.g., bitrate, latency, resilience, security) and new network capabilities trigger the need of multilayer application-aware networking.
In this Project, we form a hypothesis that it is possible to design new optimization methods for multilayer application-aware networks in order to improve the network performance by utilizing additional information that can be provided by cognitive processes including data analytics mechanisms based on machine learning methods. Potential recipients of the Project results are network operators, network service providers, and producers of network equipment and software, since results of the Project can be applied to design and optimize communication networks.
2021.02.03 Seminar: Aleksandra Knapińska “Machine-Learning Based Prediction of Multiple Types of Network Traffic”
2021.02.03 Seminar: Damian Mroziński “Optimization of conergent packet switched 5G transport networks with latency-sensitive traffic flows”
2021.02.03 Seminar: Dominik Dulas “AI-assisted dimensioning methods for Network Slicing in Next Generation Mobile Networks”
2020.12.02 Seminar: Dr Piotr Lechowicz “Regression-based fragmentation metrics in spectrally-spatially flexible optical networks”
2020.12.02 Seminar: Prof. Miroslaw Klinkowski (National Institute of Telecommunications, Warsaw) “Optimization of 5G conergent transport networks with packet switching and time sensitive traffic”
2020.10.07 Seminar: Prof. Krzysztof Walkowiak “Advanced methods for optimization of multilayer application-aware networks – Project kick-off meeting”
Cloudflare announced support for a new proposed DNS standard called Oblivious DNS-over-HTTPS (ODoH). The protocol, co-authored by engineers from Cloudflare, Apple, and Fastly, works by adding a layer of public key encryption, as well as a network proxy between clients and DoH servers such as 220.127.116.11. The combination of these two added elements guarantees that only the user has access to both the DNS messages and their own IP address at the same time. Cloudflare claims that the protocol shouldn’t cause any significant changes to browsing speed while vastly improving DNS privacy at the same time.
2020 showed the need for reliable internet services all around the world, especially with the rapid move of many aspects of people’s lives online, including work, education and healthcare. Several predictions are being made about what 2021 has in store for optical networking. <br> A very important issue is ensuring, that reliable broadband services are available to as many people as possible, including rural and underserved areas. For that reason, increased support for public funding of broadband networks is expected. Working online showed the importance of remote collaboration, videoconferencing and easy file sharing. Because of the need, new, innovative cloud-based solutions for remote work are expected to appear. Also 5G and FTTx - various methods of connecting fiber optic cables and equipment to residential homes and apartment buildings will be further developed. Since the network resources are utilized more than ever, an accelerated adaptation of high-speed connectivity solutions and increased deployment of pluggables is also forecasted. With the development of networking equipment, a concept of open optical networking is expected to also be gaining more attention. The idea is to seamlessly integrate the best-in-class elements from various suppliers into a unified network.
According to a new study, the 'streaming wars' narrative created by the press proved to be incorrect. All major streaming platforms have seen a growth in subscriber count. This is a result of users adding new subscriptions without unsubscribing the ones they already pay. For example, Netflix subscribers are taking approximately three other streaming services while HBO Max subscribers — almost five. This lead to Netflix still being the dominant service, surpassing 200 million subscribers worldwide. The numbers are expected to grow even higher, as much of the US re-enters some form of self-quarantine this winter. This will likely lead to largely increased internet traffic in the video streaming domain.
Amazon will now allow car makers to bulid their own voice assistants based on Alexa AI and use custom commands. The users will be able, for example, to ask the assistant roll down a car window, or how to troubleshoot a device without connecting their mobile phone to the car. As stated in the blog post, Fiat Chrysler Automobiles is the first Alexa Custom Assistant customer, and has already begun the planning process for the development of their intelligent assistant for integration in select vehicle models. This move might lead to increased internet traffic in the car domain.
As video consumption has seen an increase during the pandemic and people have been spending more time at home, a new way of watching video grew in popularity — watch parties. This feature enables simultaneously watching the same show by multiple users and live commenting and reacting to what is happening on the screen. It triggers interesting QoS requirements, including synchronising both the show and the chat for all the users, on top of adapting to the differences in their internet connection speeds. Hulu recently rolled out the feature for all their users and also added a "click to catch up" option — if one of the watch party participants pauses the video, while it continues playing for the others, he can later catch up to continue watching the show live with friends. <br> A similar watch party feature was introduced by Amazon Prime earlier this year, and HBO partnered with Scener to enable co-watching to their users.
Facebook enters the cloud gaming market. Users will be able to play games streamed directly streamed from Facebook’s data center without the need to download them. The idea is similar to services offered by Microsoft and Google, but without the console-quality games offered by those services. Facebook gaming will be available on their desktop website and Android app, initially only in locations close to Facebook data centres (north- and middle-west United States).