how to install hmac on the Openwrt???

[ssssaaamm]

I am interested to install hmac in OpenWRT too. How can we do this job?

[tiuman]

if you can do it install a tdma in openwrt with other software please report to me about it?? i can't find right method

[CodeFetch]

@tiuman @ssssaaamm I have tried to use HMAC on OpenWrt a while ago (I think barrier breaker). Here is the kernel patch file: https://raw.githubusercontent.com/freifunk-graviton/graviton/patch/patches/openwrt/0081-add-hmac-support.patch

Unfortunately the whole zeromq thing does not make sense using OpenWrt. Another thing is using C++ needs much flash memory as you need all the standard libs. So you end up writing the user space program in C just to see, that you need to implement the scheduling in kernel space instead of using netlink to make it work reliable on commodity routers (and the timing is still an issue).

I'm thinking about implementing something like eMCCA in 802.11s as that could become part of the mainline kernel in a distant future. All the generic power save and timing hooks needed already exist. So if you know someone who is interested to participate or wants to support it in some way leave me a message.

[Tim1Wu]

@CodeFetch Hi Do you have any latest process?Or do you have any good idea? I am most grateful for your selfless help

[CodeFetch]

@Tim1Wu I think MCCA should be implemented in ath11k. But for that one needs the source code for the firmware and Qualcomm is not willing to offer it for free. I guess it will cost $200k+ for getting it implemented. People will also have a look at Mediatek and researchers at https://github.com/open-sdr/openwifi for alternatives.

There is a "premium" feature in the ath10k firmware called QBoost which is some kind of polling-approach to TDMA and might be interesting to you. I have a patch for it, but I think there is some synchronization code missing as the performance is bad.

802.11ax gives us much better possibilities to do collision avoidance and has a better timing granularity. Extending and implementing the MCCA standard for 802.11ax is therefore what we should focus on. When we solve the "hidden node" and the "fastest node" problem with that, sophisticated routing (on top of WiFi 7) will make it perform better and after that only fractal coding together with agent chatting will come. That will possibly be another chip generation (WiFi 8) as it would profilt from AI acceleration to do the carrier switches for the different fractal patterns and doing guessing-work for predicting the choices of other nodes.

By the way... Intelligence including AI is always about predictions. My father says: "In the 80s we called it statistical analysis". What makes a human being intelligent is mimicry (mirror neurons etc.) and imagination. So what we could get from AI in collision avoidance is "imagining" the traffic flows of other devices and mimicing the behaviour of other nodes in a provision-phase. And it can get as mad and erroneous as every other intelligent organism... Concerning the additional power consumption needed for calculating these predictions, the limiting factor of "knowledge" which itself must either be transferred or extrapolated and the user expectation of comprehensible behavior I don't see this coming before 2040 which might bring us a not-so-much-AI WiFi 8 first in 2030.

It's forseeable where this journey will go until we reach "physical spectrum" throughput. I'm working on some of these things already, but getting funding is hard. So I play my role as a consultant at the moment.

[Tim1Wu]

@CodeFetch Thanks for your answer, but I am pessimistic about the research prospects of this project.By the way, could you please send me the QBoost patch as well as the related pictures, information or anything else you mentioned? I would appreciate it very much.My Email wuguodong@outwitcom.com