kb/s speeds

support

#1

What speeds are people getting showing on there apps? I am only getting around 27 kb/s and seems very choppy. I am running all the latest software and on my iphone x. Some of the cameras are only a couple feet from my router. Not sure of the 2.4 up speed, not in front of it right now, but I have a very fast connection for everything else.


#2

Are you in HD or SD mode? The upload speed will vary in either case depending on how much compression can be done, which depends on the amount of motion going on, the lighting, and the amount of variation in the scene.


#3

I tested it both ways. I get about the same for both…it bounces form about 27 to 42 and back.


#4

Can you run speedtest.net while connected to the same 2.4GHz wifi and see what the upload speed reports? Do you have any other devices on the network that might be using upstream bandwidth?


#5

Looking at my logs on speedtest, I am ranging from 14-21 mbps up on the 2.4


#6

Sounds to me like you have an old (as in more than 2 or 3 years old) WiFi Router. I recently upgraded my Linksys WRT54GL to a new Netgear Orbi RBK22 Mesh Router (available exclusively at Costco for $199.99). My WiFi speeds went from 27Kbs to 117Kbs+.


#7

I have an ASUS AC1750.


#8

Ok, probably not your router. But symptoms sound like a wifi issue. Very likely your phone is using the uncongested 5GHz band. The Wyze cam’s only work on 2.4GHz. I don’t know what “wifi analyzer” apps are available for iPhone. On my Android I have WiFiAnalyzer (open-source) VREM Software Development. It can show what channels are least crowded. I run Tomato shibby firmware on my ASUS (very old N16). It has the ability to do a “wireless survey”, where it enters listen only mode and displays channels, and displays RSSI (Received Signal Strength Indication), Noise, and Quality (difference between RSSI and Noise).

You may be able to find a better channel. I am not sure how the ASUS firmware works (I don’t use it). Many do a scan at the time they are powered on, and choose at that time. Also, I recommend using 20MHz (not 40) channel width on the 2.4GHz band (unless you live out in the country with no close neighbors)

Another thing to consider: Evenings are a bad time for wifi (because so many people stream video to their TV over wifi, and that is when they are most likely to watch)

What version of camera and firmware? Is it easy to move the camera to see if that makes any difference?

Wifi is convenient, but that’s about its only upside in my opinion. Especially for a fixed location device that already has a wire going to it. But it is the only option you have with the Wyze cameras, and with the current chips, you are limited to the 2.4GHz band which is really only 3 non-overlapping channels (1, 6, 11). But if you are in an apartment, or even a residential location other close wifi routers, there will probably be wifi APs camping on channels between. So choosing 1, 6, or 11 doesn’t protect you from other APs that are using the “5MHz wide” channels between.

 


#9

I’ve been guessing that the KB/s number is the rate that data, bytes, are being received by the app. I’ve also been guessing that the images are compressed plus the only information sent is what is necessary to update the pixels in the image that have changed. So if the camera sees a scene that is very stable it needs to send very little data and the KB/s will be very small. For instance a camera in night mode off in a completely dark room would send maybe 5 KB/s. A camera looking at trees in a strong wind would send 100+ KB/s. A more interesting number is the number of frames/second the app is receiving. I believe the rate the Wyze cameras run is 10 frames/second. The app, at least android, doesn’t show a frames/second number. The TinyCam app does and when the video gets choppy I see the frames/second drop down. I’m guessing that if the network connection and speed are adequate, the app receives an update to the images 10 times/second. Otherwise if you are receiving frames at something slower than that, some frames are lost, you see choppy video.

I’m not a Wyze person, just someone having fun with these cameras.

 


#10

That was my feeling too, that the metric is how much has been downloaded as changes between frames.


#11

@Kit and @oaktree - both good posts. Yes, it is very dependent on how much has changed. So my last post was probably not very relevant to the issue.


#12

Stream speed? is it Kilo Bytes? or Kilo bits?

The iOS app is kb/s lowercase, the Android app is KB/s upper case

I’m seeing about 100kb/s in iOS and same value in Android 100KB/s

It’s probably Kilo Bytes which is around 1 mega bit stream speed for an HD video

Can anyone confirm? Kilo Bytes or Kilo bits?


#13

I am reasonably sure it is KiloBytes/s. But I measured more than the App displayed.

I have a router running Tomato Shibby, and it can display traffic per ip address. I connected from my Android phone to a v2 camera in a dark room (10 miles away). When I am viewing, the Android display jumps around between 37 KB/s and 70 KB/s. The “real time” graph shows that it is using 850 kb/s or 104 KB/s. (this is what is being received from internet and sent to the Android. The v1 camera at same location Android app claims 30 - 45 KB/s and Shibby reports 640 kb/s or 80 KB/s.

So I am not sure what the Android is reporting. Perhaps video stream speed not including any “encapsulation” (including IP. TCP and whatever overhead the streaming protocol Wyze/Xiaomi is using).

So it appears to stream to another location, your internet upload speed must be able to support at least those values, where nothing is going on and video can compress very well. If there is movement, then more “information” needs to be sent, with higher bandwidth.

This is one reason that for users with DSL, the hope of streaming multiple cameras simultaneously to the App to be displayed in multiple camera view is probably a pipe dream. More doable from a bandwidth standpoint would be local streaming to some type of a “hub” that could then process the video streams into a multicamera view and re-encode as if it were a “virtual camera” ideally using a wired connection to the router.

I think that many home wifi would also be pushed to stream too many (more than 4?) cameras simultaneously. Wifi shares time, and if you have one camera that has a poor signal, it uses a lot more air time to get the same data through, so a slow device can have a detrimental effect on other devices using the same channel group (each 2.4 GHz channel is only 5MHz wide, so for example, using 20MHz channel width centered at channel 6 uses up all of 5 & 7 and half of 4 & 8. So there are really only 3 “channel groups” available in 2.4GHz. 1, 6, 11.


#14

I appreciate the time you took to check this out. It makes me even more curious


#15
This is one reason that for users with DSL, the hope of streaming multiple cameras simultaneously to the App to be displayed in multiple camera view is probably a pipe dream. More doable from a bandwidth standpoint would be local streaming to some type of a “hub” that could then process the video streams into a multicamera view and re-encode as if it were a “virtual camera” ideally using a wired connection to the router.
I'd like to see a multi-cam view with a user-adjustable frame rate. So you could set the frame rate to 1fps or even one frame every 2-5 seconds. This way, you would have an idea what's in front of each camera before actually switching to that camera for live view.

#16

Yes, it would be nice to have a “thumbnail” view of multiple cameras. But I am worried that at some point, the resources on the SOC (RAM, processing) are going to be exhausted. We are already asking it to do “detection zones”, and once that was provided, the requests just jumped to the next level, e.g. “multiple detection zones”, “exclusion zones”, “polygon zones”. “arbitrary mask layer”. These can all be done given enough memory and processor, but my guess is that the SOC is pretty limited in what resources it has. It isn’t the same processor you would find in a high end iPhone.


#17

I don’t see how the capabilities (or lack thereof) of the camera’s processor/ram would affect this. If each camera is capable of streaming 10-15 fps while performing its other functions, then surely it could stream 1fps or less. I would think that if anything, the impact on the local wifi network might be more of an issue in a multi camera environment.


#18

But what if you are recording to SD? Then the chip has to be processing the video stream for that operation. If it is streaming I am guessing that it would be streaming the same “processed” video as it is recording.

If that is a valid assumption (may not be), then having the camera also create a 1 fps stream could require a different video buffer, etc. These cameras probably have ~ 64 MB of RAM (still 100 times what a maxed out 1982 IBM PC had), but no where near what a mobile phone has (1 GB is a low end device). I remember when we upgraded our VAX 11/780 from 2 MB to 8MB in 1984 and it cost over $20K for the upgrade… showing my age here.

Perhaps the SOC has more resources than I am giving it credit for. I just don’t know. The v2 camera supposedly uses the Ingenic T20 SOC

https://www.reddit.com/r/wyzecam/comments/8b8d79/hardware_mods/

Here is a “data sheet” for the processor.

ftp://ftp.ingenic.com/SOC/T20/T20_PB.PDF

It appears there are two parts, one with 64MB (512Mb) and one with 128MB (1Gb) of integrated DDR2 RAM. On a cost constrained device, we can assume on the low side. Edit: just saw this in the “Specs” for v2 camera, so my assumption was incorrect if the specs are correct.

CPU 1.0GHz
Memory 128MB

I doubt the v1 camera has more.


#19

That’s a good point about still having to maintain the mSD card recording while also streaming at a lower rate. I have no idea if that would be a show stopping burden.Ssomehow I would think that extracting one out of every 10 or 15 frames would not be too processor/memory intensive. But who knows? Actually, the Wyze engineers know. It will be interesting to see if they are eventually able to produce something like this.