by Volkan Sevindik
Traffic classification, or differentiation, means creating different traffic groups based on the service quality needs of the generated traffic. Traffic classification is mainly used for charging purposes. Network providers want to charge users based on the service they are consuming from their networks. Users usually subscribe to the traffic class(es) through their service level agreement (SLA) made at the time of their service subscription.
Network providers promise to deliver the services that users sign-up for in their SLAs. Failing to deliver the promised level of service decreases the overall customer satisfaction.
In this article, I will use two terms for the “network” users are connected to. The traditional big cellular telecommunication network is called macrocell since the network covers large geographical areas and serves a large number of users. The term, femtocell, is used both for the geographical area that is served by the femto base station (femto access point), and for the femto base station itself. The network area served by femtocells is known as femtozone. Most of the time, network operators refer to the femto base station as femtocell, and I will do the same in this article.
Network providers should be able to deliver the same Quality of Service (QoS) and Quality of Experience (QoE) regardless of the network that users are currently connected to. QoS means ensuring the promised service levels at all times, regardless of the resource limitations in the network. QoS is usually used as a term to determine the quality requirements of the delivered service. On the user side, QoE is the widely used term to measure the quality of service experienced by the users in the network. One can think of QoE as a projection of QoS (network side) onto the user domain. Users may request services from a macrocell or femtocell depending on whether they are outside or inside a building. That is, the serving network differs as users change their geographical location. When a user hands off (switching the serving network) from a macrocell network to a femtocell network, femtocells have to distinguish the traffic class of the user and start providing the same QoS level ((high throughput (video), low delay (VoIP)) which is specified in user’s SLA. Therefore, we require traffic classification in femtocell environments (femtozone) for usersʼ QoS and QoE continuity. Users requesting high definition video, high definition voice or high definition video conferencing should receive the same service without any QoS interruption in femtozone.
Business femtocells are expected to serve more users than home femtocells do. So, there may be a bottleneck when large number of users request high class traffic from the femtocell. One key point to note, increasing demand for HD of any sort (HDOA) will require capacity in a heightened way that even femtocells will not be able to handle. Some examples of HDOA traffic are listed below;
1- HD Presentations and HD Video upload/download for marketing department
2- HD Excel for finance and accounting department
3- HD remote desktop and HD file transfer service (FTP) for engineers
4- HD voice for management department
5- HD email for human resources department
6- HD GPS for supply chain and operations department, in order to track shipped and received products with high accuracy
Depending on their needs, one department can receive one or many of these services. The different services need to be delivered by business femtocells, and network providers should be able to configure femtocells remotely, or businesses are given an account through which they can configure the supported traffic classes in their femtozone.
Accounts are created in the policy server which locates in service operators’ core network. At the time of femtocell purchase, businesses will be given an account to login to the server and change the configurations of their femtocells. Traffic classifications created by the businesses may create some bottleneck in the backhaul (DSL connection), if large numbers of high class traffic users are entered into the femtozone.
Some users may only request HD video conferencing, while others may need HD voice, HD document (means that document response time is shorter) and HD web service at the same time. One may argue that femtocells are developed to provide high quality of service anyway, and there is not any bottleneck in the backhaul part of the network. However, the number of users and the number of HD traffic types will be increasing in the near future. Therefore, resource management will be a challenging problem for femtocell networks like it is for macrocell networks.
Download this Article in a PDF Format
About the Author:
Volkan Sevindik holds a BS degree in Electronics and Telecommunication Engineering with a first-class honor distinction from Istanbul Technical University, MS degree in Electronics and Electrical Engineering from Bogazici University, Istanbul, Turkey, and a PhD degree in Electrical and Computer Engineering, University of Massachusetts, USA. Sevindik is a Systems Engineer at Airvana, Inc., as well as a Technology Expert at LTE Portal Broadband Solutions, Inc. He is deeply interested in next-generation wireless networking technologies with a special focus on new quality of service (QoS) and quality of experience (QoE) models. His current interests are admission control, scheduler design, traffic classification, radio resource management in cellular telecommunication networks. Sevindik often blogs on LTE Portal, as well as his own blog at www.volkansevindik.com.
Popular Posts
Vladan Jevremovic
Volkan,
On behalf of us at iBwave, I want to thank you for your insightful article. Your entry is the first of (we hope) many blog entries written by in-building industry experts.
I have few questions & a comment:
In your experience, what are the typical values for latency (delay) for HD vs. non-HD data applications such as the ones meniotned in the article (HD Excel, HD file sharing, HD presentations). How do they compare with typical HD voice delay requirements?
HD service requests impact maximum user capacity per femtocell. More HD service requests – less user capacity per femtocell. Given a fixed number of users in a femtozone, we can say that the number of femtocells increases as the number of HD service requests increases.
However, the more femtocells we have in a femtozone, the more inter-cell-interference those femtocells generate. High interference reduces signal to intereference and noise ratio (SINR). Once SINR drops below a critical value, mobile may not be able to transmit and/or receive the signal due to poor SINR. Therefore, poor SINR affects QoS and consequently affects QoE as well.
To aviod femtocell “overpopulation”, RF simulations should be done in femtozone to determine SINR levels and compare them to the benchmark SINR levels needed to maintain HD transmission of all the HD services requested by the customer.
Regards,
Vladan Jevremovic
Volkan Sevindik
Hi Vladan,
You are very welcome! I am very happy to contribute to the blog. Thank you for your question and comments.
Please find my answers below;
For video application, networks are expected to deliver less than 250 msec. delay. That is, maximum end-to-end delay should be maximum 250 msec., (delays less than 200msec. is recommended for better QoE). The basic difference between standard video and the high definition video is the bit content carried by frame. HD video bit rate changes from 2.6Mbps to 8Mbps depending on the Codec. Even though the same delay performance is expected for standard video and HD video, the bit rate requirements are very different.
It is true that increasing number of video users will require more average sector throughput, which is defined as the total traffic transmitted from the femtocell (assuming femtocell only has one sector). And more femtocell deployments will be needed to satisfy the demand of HD video users. This will not create a problem if femtocells use different frequency channels. If femtocells are deployed using the same frequency channel as the macrocell, self organizing network (SON) algorithm will decrease the coverage area of the femtocell in order to decrease the interference between femtocells, and interference on the macrocell. This will result in less coverage and high capacity network (basic dilemma in macrocell network design). What I recommend is to use FGW as a central radio resource management unit to manage the resources based on geographic areas, rather than on a single femtocell-by-femtocell basis.
Also, soft interference canceling (SIC) techniques will also help in reducing the interference. These techniques are used by almost every LTE femtocell.
One important aspect of video delivery is the scheduling process. Scheduling algorithm used in femtocell will optimize the video delivery by taking into account delay, throughput, service level agreements, QoS levels of a particular user. If there are standard video users in the coverage area of femtocell, scheduling algorithm will make sure that those users will also receive their video together with HD video users.
So, for a delivery of HD or SD video, you need;
1- Scheduling algorithm for optimum video delivery
2- Backhaul that satisfy throughput and delay requirements
3- Separate, clean (less interference) frequency channels
4- Fast SON algorithm to select optimum carriers based on interference levels.
These four items are important for a successful video delivery.
Please let me know if you have any questions, or comments.
Thank you,
Volkan