mdapa I don't know if this is helpful, but all of the networks I have used recently are from my university.

This may well be relevant. Enterprise-quality access points are different from regular access points. For example, they often cooperate behind the scenes (over a wired network) in terms of how specific access points behave toward specific endpoint nodes. For example, if the collective wisdom of the access points is that you should be using #4 instead of #11, but you are signed in to #11, #11 might either drop you or slowly degrade service to you to encourage you to switch.

Meanwhile, some enterprise-quality access points behave differently toward different client devices -- for example, if the AP chipset was made by the same company as the client-node chipset, private protocol extensions may be used.

So it might be interesting to try Wi-Fi networks from a different provider -- maybe a public library. In the U.S., I think the McDonald's chain uses enterprise-grade Wi-Fi gear and allows public access. Also, if you can find out whose equipment your university Wi-Fi uses, that could be interesting too.

      de0u I have been trying some methodical experiments. Using analiti to analyze my connection I have connected to both my school's network and my hotspot using every network privacy option.

      I tested each option for around 2 minutes. In the app I selected "continuous download traffic throughput & bandwidth" since I've noticed the disconnections mostly occur when my phone is actively using the network.

      When connected to my university's network, it made no difference if I had per-connection, per-network, or device MAC. Every 30-60 seconds my phone would disconnect from the wifi then reconnect after a few seconds. This meant over the 2 minutes it always disconnected twice.

      When connected to my hotspot (which itself is my old phone that is connected to the university's network) it again made no difference if I had per-connection, per-network, or device MAC. My phone never disconnected from the wifi over the 2 minutes.

      analit does give some info on the type of access point. It says that my room's access point is by "Hewlett Packard Enterprise". I don't know if this is consistent across the entire campus but every one that I've seen at least looks the same as this one.

      It might be a few days until I can test another provider's network. I'll report the results of that when I do. Let me know if there is anything else I should try in the meantime.

          mdapa It might be a few days until I can test another provider's network. I'll report the results of that when I do. Let me know if there is anything else I should try in the meantime.

          So far this is a good report. When you issue your "final report", please re-state that you believe you did not encounter these issues with the stock OS.

          You might try asking your university's tech-support team if they have had other disconnection reports and/or if they are willing to tell you why your device is being disconnected. Somebody would need to hunt through the AP logs a bit, but there may be interesting information there.

          FWIW, I am not aware of "HP Enterprise" being top-ranked among enterprise access points. This randomly-selected web page suggests that HPE bought Aruba, which previously I would have classified as reasonable. Meanwhile, enterprise access points have many options, so it's possible your university has turned on some feature that others don't.

          Wi-Fi is complicated.

            To add to the above, using the same WiFi network as yesterday, I'm now seemingly not getting dropouts with MAC randomisation set to any of the three settings. I'll update if Android Auto is the same (so far that has consistently dropped out every minute every time I've tried), but it would appear whatever the bug is, it can be inconsistent.

              I'm also now going back to wondering if it's frequency related. Most of my testing before was on a joint 2.4/5GHz network, so I would have been on 5GHz. My wireless Android Auto box is also 5GHz-only. So it might be worth disabling 2.4GHz/6GHz on a network and seeing if it has the same issue.

              mdapa any chance your university network is 5GHz and your hotspot 2.4GHz?

                  Gilboboy Please report back if your Android Auto wireless is working. If Iam connected I have no possibility to change the WLAN Settings from Android Auto

                      kebapfan I can confirm my Android Auto is still not working and disconnects every minute. I'll test later whether this is some kind of special case or 5GHz networks.

                        After further testing, I'm almost certain this is related to the 5GHz band. By changing my home WiFi network to be 5GHz-only, the disconnects happened every minute, with it set to 2.4GHz/6GHz, no disconnects occur.

                        My guess is the inconsistency we've seen is because if you have an access point broadcasting multiple frequency bands, it'd depend which one you were in as to whether or not you get disconnected.

                        This would explain why wireless Android Auto always disconnects, because it forces 5GHz. If we also assume that mdapa's university enterprise gear was 5GHz and their old phone hotspot was 2.4GHz then that's also explained.

                        MAC settings also made no difference to whether I was disconnected or not, the frequency band was the only thing that changed the outcome.

                        I hope the GrapheneOS Team can have a look at this

                        Gilboboy I just checked and this is the case. My uni's wifi is 5GHz and my mobile hotspot is 2.4GHz.

                          Gilboboy I've submitted an issue here: https://github.com/GrapheneOS/os-issue-tracker/issues/3944

                          That seems likely to be upstream.

                          Please feel free to subscribe to the issue or use a reaction (e.g., thumbs-up), but for previous GitHub issues the developers have firmly requested that people not make "me too!" postings -- in the past that has resulted in the developers just locking comments on an issue.

                            Good suggestion.

                            For anybody that it's helpful for, I've found that it's specific 5GHz channels that are affected. If you are able to change the 5GHz channels your access point uses, you can find one that doesn't disconnect for you (my Android Auto wireless hasn't disconnected when using Channel 165).

                              Gilboboy For anybody that it's helpful for, I've found that it's specific 5GHz channels that are affected.

                              This is great information for people to have. Thanks for reporting it!

                                Gilboboy where can you Change that? If Iam connected to Android Auto Iam Not able to change anything with the WLAN Iam connected to

                                  Gilboboy
                                  If thats an actual issue, then that to me would suggest a limitation of the wifi chip in the some of the Pixel devices..? I havent had any disconnection issue at home, but my access point is set to chanel 36, 80mhz channel width & 802.11a/n/ac mixed.

                                    Were all these Pixels purchased through different vendors or through 1 vendor?

                                    Could thr phones have bren tampered with prior to arrival because they want cell service used?

                                      Ive had the same problem on my Pixel. My solution so far has been to change my Routers (Fritzbox) 5Ghz connecting channel to 140. Connecting to the lower channels 38, 40, 44 and so on has caused frequent drops with no reconnect. I havent tried all channels yet, but fyi im the only one using this channel in my wifi range. Also using "per-network randomized MAC". Dont know whats causing the problem, but this has been working for 1 week now without issues. Hope this helps someone.