We are exploring using NetSim to study QoS in 5G for video application. Briefly, we seek to modify network parameters like UE-gNB distance, mobility, pathloss, transmit power, network congestion etc and study impact on throughput and delay. Looking for some good suggestions and references. Thank you.
There are two options. You can use the “synthetic” video model within NetSim. Just set up the network layout within NetSim using drag/drop. And then set the video traffic flow (source to destination). If you want to send a “real” video file. Then NetSim interfaces with the FFMPEG program for handling video files at the sender and the receiver. For both cases, you can then run simulations by modifying various parameters and analyze the results.
We are looking at mobile nodes. Both FR1 (< 6 GHz) and FR2 (mmWave) are being considered.
For sub-6 (FR1) you can assume distances, transmit powers and path losses similar to LTE (since almost all current deployments are n78 and collocated with LTE basestations).
For mmW the distances will be on the order of 250 m, and the path losses much higher. There are several published works (e.g., "Propagation Path Loss Models for 5G Urban Microand Macro-Cellular Scenarios" from Qualcomm, "Path loss models for 5G millimeter wave propagation channels in urban microcells").
Regarding network congestion, the sub-6 may have ORAN-split fronthaul which will never be congested, the F1 interface will be similar to the backhaul characteristics of 4G (i.e., very infrequent congestion due to MPLS-TE or CE), but then the source may sit in a DC or be connected over the Internet. For mmW there will probably not be FH - only F1. The data rates will be higher, but in many cases well engineered links. Regarding the last segment there will be variability. Some operators will attempt to ensure good QoS (to justify a premium payment for 5G) and employ CDNs, etc. Some might just say that form their handoff points it's not their problem, and since the rates may be high the performance may be problematic. Expect uplink video to be even worse in such cases.
Yaakov J Stein Wouldn’t MIMO, Beamforming, different numerology, and all the new 5G technologies give much better performance even in FR1 when compared to LTE? Please also let us know your comments on the FR2 RAN performance?
The factor of 10-100 increase in data rates of 5G as compared to 4G is mostly from larger system bandwidth (maybe a factor of 3 can be obtained at the same bandwidth). These higher than 20 MHz bandwidths are mostly available in FR2. Yes, the C-band allocations technically support 100 MHz and u=01,2, but the question is whether there is enough free space for a particular operator (or private CBRS user).
MIMO, whether used to exploit spatial diversity and thus enable frequency reuse, or for beamforming and thus higher effective SNR, indeed enables higher data rates. But it requires accurate beam-tracking for mobile users, and expensive arrays even for FWA (64R64T and 128R128T already available).
A lot has been said about the mmW challenges in the air interface (higher free-space attenuation, absorption, amplifier efficiency, health issues, etc.) but there are major issues in the xhaul as well (see the new book - "5G RAN, the dark side of 5G"). The new single-lane 100G standard (4-lane 400G) is a major step in the right direction, but the cost of the optics is still prohibitive.
The major issue right now is that the supposed 5G deployments are really only collocated gNb for eMBB using option 3 (or 3A or 3X). This allows quick upgrade, but quickly saturates the xHaul network. What is needed is densification, but stretching new fiber access to 10-100 times the number of sites is going to take a while if it ever happens. In some cases IAB midhauling to the hub site will be great, but if the operator doesn't have mmW bands not being used, that will require sharing bandwidth between the users and the midhaul.
Wondering what is going to be better in terms of cost and mobile battery duration:
- upgrading from 4G to 5G
- upgrade to a more efficient (new) video codec that requires, e.g. half bandwidth for the same video quality.
David,
Using higher system bandwidth with the same transmitted dB/Hz obviously scales the battery consumption. The larger FFT sizes also scale cost and battery, but modern SoC processes reduce power usage to offset this.
The video compression ratio won't have a major effect on cost for small devices (phones and tablets), but might impact FWA large screen devices with HD and higher resolutions.
Yaakov J Stein I am interesting in know what may happen to performance during a handover? Any good references? Thanks
You can see these slides: https://tetcos.com/pdf/Performance-analysis-of-5G-Handovers-using-NetSim.pdf
I had some questions since I am working on something similar. What are the different codecs that are supported? Can the video rate be changed? How is performance measured?
The codecs supported are H.261, H.263, MPEG1 Low Res, MPEG1 High Res, MPEG2 Low Res, MPEG2 High Res. Additionally, Buffered Video Streaming Models Options BV1 through BV6 are also available.
Yes, the video model input parameters can be changed to alter the video rate.
Performance measurement is common for all applications. The results output includes time averaged throughput, latency, error, jitter, etc., and then time series output (metric vs. time) is also available these measures.
- Badges
- Science topic
- Similar topics
- Philosophy
- Moral Philosophy
- Free Will