I ran across Praveens blog https://praveenkumar4blog.wordpress.com/ as was trying to recall the terminology for aggregate frames. I almost have the name and details of each layer, MSDU and MPDU. Praveen has some good details on Fortinet it appears. It has been years since I have looked at a Fortinet product, they are into all sorts of products now. Ahh Fortinet, I never hung out with you much. . . maybe I should now? Years ago they only sold firewalls.
The whole concept of MSDU, a service data unit, is not clear to me. Reading about it, I pulled up the IEEE specification document, IEEE Std 802.11-2016. If you want to understand how these things work, read this document. You have to register with an email address on IEEE site, and then you can download it for free (you could a few months ago). Not all the documents from IEEE are free to download, make note of that. If you can get that document, it’s a great study aid for learning 802.11 wireless. The document covers directional multi-gigabit wireless concepts and terminology. It also is one of the only places that goes into detail on how DL-MU-MIMO) actually works.
Looking up PTK and the details, a few things to note. More new things, that I found. For MESH access points, or other equipment, you may find a thing called and AP PeerKey. It’s a protocol for access points to communicate to each other. From the document: “The AP PeerKey protocol provides session identification and creation of an AP PeerKey association to
provide for security of OBSS management communication between two APs. The result of a successful run of the AP PeerKey protocol is an AP PeerKey association. An AP PeerKey association is composed of a mesh PMKSA and a mesh TKSA.
I am still learning here about wireless frames, a few things I found out. In some of the IEEE docs, they call A-MSDU a “payload protected (PP) aggregate medium access control (MAC) service data unit. Wow, that is alot to remember, I forget the terms A-MSDU. What might help is remembering what they mean.
I do have a few questions. What causes an AP or station to implement aggregate frames? Is it basically a queue, and if the queue begins to fill, and there are no transmit opportunities, then “start” and begin building a MSDU frame? I am unsure.
Question, how does a PPDU relate to a MPDU? And how does that relate to service period (sp)? Answer- I think the PPDU can be anything transmitted, like a beacon, or a packet. In the ieee doc, we see on page 2325, “PPDU is formed during data transmission by appending the PSDU to the Exteneded Rate PHY preamble and header. At the receiver, the PHY preamble and header are processed to aid in the demodulation and delivery of the PSDU.” And the service period? a time you are told to transmit or assume you can transmit. My definition, anyway.
How does a TIM (traffic indication map) differ from a beacon?
What is a TSID?
Things change with MIMO. A few things. “Very high throughput (VHT) multi-user (MU) physical layer (PHY) protocol data unit (PPDU): A
VHT PPDU with a format that is capable of carrying up to four PHY service data units (PSDUs) for up to four users and is transmitted using the downlink multi-user multiple input, multiple output (DL-MU-MIMO) technique.” thats from the document. So in MIMO with multiple users on a access point, it reads to me like the AP may have a PPDU to send that has data for up to four (4) different clients. Might make for an interesting conversation with speed uploads. If you have a busy wireless network, using MIMO, your upload speed maybe slower than download speeds. Rationale? With MIMO, downloads are more efficient to clients, from the access point. More effective use of air-time, and uploads have no such MIMO capabilities in some STA. Its the downlink that is faster (effectively faster, actually bits per second, it is the same speed. . .it is the same bits per second!)
Another question, is an MPDU with many MSDU always a good thing? I like to think most designs have good and bad. So when the concept of packing frames together into longer frames, whats the catch? Yes there is a downside to most designs. For frame aggregation, mostly all good things. But can it cause latency? I would say yes, even if it is a 1 millisecond latency, it still maybe the case. Latency of one millisecond is not a big deal, right? Well it is something to keep in mind. Latency here and there can add up. Your server latency, WAN latency, then wireless latency. . . speed is king. Yes speed is king, but you can be fast AND have latency. They call it a long, fat, network. https://en.wikipedia.org/wiki/Bandwidth-delay_product I can see how this make wireless more effective, in general. But at times, could this cause latency issues? Or does this only happen when the signal level is very good (high speed networking)?? If signal level (RSSI) is poor, speed is low, does protocols still try to pack the data using these process? If so, latency will be impacted.
Did you read the link on wikipedia? if not, click the link, its in the previous paragraph. What did you see? From that link, it says this:
A network with a large bandwidth-delay product is commonly known as a long fat network (shortened to LFN). As defined in RFC 1072, a network is considered an LFN if its bandwidth-delay product is significantly larger than 105 bits (12,500 bytes).
So, say you have 300 mpbs wifi link, that is megabits, and that is fast. . . . . . which is really maybe around 20,000,000 megabytes per second.
What does a 1 millisecond delay cause? In one millisecond, you divide 20,000,000 divide by 1000, as 1000 milliseconds per second,
and you could have transmitted 20,000 bytes! Even though fast, your 300 mbps link is not a true speed demon, its a LFN. So you now have a LFN (long fat network) simply by having a 1 millisecond delay.
Back to PPDU on the airwaves. So within the PPDU, is MSDU, at least one. So the way I understand, these MSDU and MPDU only apply in wireless realm. Both sta have to have the HT (high throuputoptionimplemented) so that each sta knows what to do with the msdu packets. As I understand the packets are similiar to like a multiplex circuit (like LACP). After the data is recieved it is put back on the network in the original state, so the packet size goes back to 1500 or so. So there is no fragmentation.
Also the packets all need the same qos value or they won’t get packaged in a MPDU. So different qos packets should never be in a MPDU. That how I read the 802.11-2016 IEEE specifications.
Lastly, remember that all the MSDU and MPDU need encrypted before transmission. How does that happen? It appears each MSDU should be encrypted separately, then packed together. Really? What I found interesting, there is a timer in each MSDU. The timer is the lifetime of the encryption protocol. . . I believe.
“The expiration of the A-MSDU lifetime timer occurs only when the lifetime timer of all of the constituent
MSDUs of the A-MSDU have expired.” page 1365. This is the PTK lifetime timer i believe (only thing that matches that in the ieee document. So the MSDU, each one maybe encrypted differently with a different timer. . .and the MPDU will transmit MSDU with expired PTK lifetimes. . . potentially sending packets that can not be unencrypted? It seem it maybe the case.
Copyright 2020 Rod Deluhery