In WiFi, does a higher data rate imply more number of collisions?

[Varun Reddy]

I'm implementing the IEEE 802.11af protocol, which is essentially IEEE 802.11ac at a lower frequency (300MHz). There is a single AP catering to 75 users, and the server is connected to the AP by a pointTopoint link. With a ConstantRateWifiManager and a DataMode set to VHT-MCS0 (about 10.7Mbps), I see that the server receives all 75 requests from the clients. When the DataMode is set to VHT-MCS9 (about 142Mbps), the server receives only 26 of the requests from the clients. Note that all the users are within range of the AP.

I thought that this might be due to an increased number of collisions (owing to higher data rates) and attempted to modify the following parameters:
1. ns3::WifiMacQueue::MaxPacketNumber (The queue size) - I set this value to 500, and there was no change in the results.
2. ns3::WifiRemoteStationManager::MaxSsrc (Number of Retransmissions allowed for an RTS packet) - I set this value to 10000, and there was still no improvement.
3. ns3::WifiRemoteStationManager::MaxSlrc (Number of Retransmissions allowed for a DATA packet) - I set this value to 10000, and the server now received 27 requests.

Despite these changes (although they are impractical values) I cannot seem to understand why all 75 clients are not hearing back from the server in the case of VHT-MCS9. And what puzzles me even more is how the 'Data Rate' is causing this effect.
 
I have attached the logs and codes; there are changes that have been made to wifi-phy.cc and wifi-mode.cc (number of OFDM subcarriers, symbol duration and a few other parameters).

[Tommaso Pecorella]

Collision no.

Error rate yes.

Just check that all the nodes are *for real* in radio range even at VHT-MCS9.

T.

[Varun Reddy]

So I'm just a bit curious. Why wouldn't the number of collisions increase? Is that because the packet length increases proportionally with the data rate? (The number of packets would then remain the same). But if this were the case, does that mean every WiFi standard does not only prescribe a fixed data rate, but also a fixed packet (or frame) size?

[Tommaso Pecorella]

Higher data rates are achieved by transmitting a larger number of bytes in the same time - usually you just use less time to transmit a packet.

As a consequence, the time a packet needs to be transmitted is shorter, and there are less collisions.

The frame size has nothing to do with the data rate. What changes is how bytes are transmitted (the modulation and coding scheme).

Books... or wikipedia.

T.

[Varun Reddy]

Oh I see. I will read about it in more detail.

And yes you're correct, the nodes are not in range in the case of VHT-MCS9. I guess the propagation loss is higher for some reason. I increased the Transmit Gain (on all the clients) and all 75 clients received data from the server. I'll try to see where the error has crept in.

[Rediet]

Hello Varun,

Actually, like Tommaso implied, the difference lies is in the SNR required by the MCS to achieve a given PER (or the PER that can be achieved considering a given SNR). The propagation loss remains the same as long as the stations are at the same distance. By increasing the Tx gain, you have just increased the received SNR, thus making the MCS viable.

Rediet

[Varun Reddy]

So that means a higher MCS (a higher data rate) requires more transmit power?

Varun

[Rediet]

No, rather than, at equal transmit power, it requires higher SNR, thus for the receiving station to be closer. Increasing the transmit power will raise other issues (power consumption, regional constraints, interference caused to neighboring BSSs, etc). Selecting the appropriate MCS in Wi-Fi is all about range vs data-rate.

[Varun Reddy]

Yes, I was wondering exactly the same! Increasing the Transmit Power would only result in more energy consumption and interference. So selecting the right MCS is all about the tradeoff between range and datarate.

Thank you both for elucidating this concept!

Varun

[Adrian Sagario]

Hi Varun,

Me and my thesis mates are also having a research based on IEEE 802.11af standard using NS-3 simulator. And we are currently stuck when building a certain wireless example in NS-3.29. Pictures attached below for the errors shown. We are hoping for your kind help. Thank you.

-Adrian

[ever shine]

I am working on a research project and project required simulation of IEEE 802.11af standard. I see all the available wifi standards in ns-3 but couldn't find the previously mentioned standard. I've searched online and found many authors mentioned in their research paper that they have implemented 802.11af in ns-3. Please guide me how can I get 802.11af ns-3 implementation.

[Tom Henderson]

On 9/1/19 4:42 AM, ever shine wrote:
> I am working on a research project and project required simulation
> of IEEE 802.11af standard. I see all the available wifi standards in
> ns-3 but couldn't find the previously mentioned standard. I've searched
> online and found many authors mentioned in their research paper that
> they have implemented 802.11af in ns-3. Please guide me how can I get
> 802.11af ns-3 implementation.

Sorry, I'm not aware of an 802.11af ns-3 implementation. Perhaps you
can approximate it by changing the channel widths on the 11ac
implementation.

- Tom