When there are many smart devices at home, video calls get blurry, and the lag is upstream bandwidth, not 300M broadband running out

Last autumn, I added two cameras and a video doorbell to my home, plus the original sweeper, door lock, curtain motor, and two smart speakers. The Mijia app on my phone has been counted to fifteen. It was from that time that my mom started complaining during video calls with me, "Why is your screen blurred like mosaics?" I started Tencent Meeting myself, and every ten minutes I shared my screen would freeze. 300M broadband shouldn't be performing like this.

I initially suspected it was WiFi. I tried changing the channel and maxing out signal strength, but it didn't work. One night, I simply squatted next to the router to refresh the background app, this time not looking at anything else, just focusing on the "Speed" column on the device management page. After watching for nearly half an hour and reconciling the process and accounts, the issue was so clear it surprised me: the downside was indeed very short, while the upstream was blocked. This article will cover the entire story from beginning to end—how I figured out who was eating in the backend, how I settled the score, and which three places I actually made. My home uses a Mercury machine, and the process of Tenda is almost the same. I'll mention the interface issues as well.

First, understand one thing: the direction smart devices are working is the opposite of what you think

.

When watching shows or scrolling on your phone, data is being fed from outside into your home—that's downhill. The "300M broadband" provided by the operator also refers to downlink. But smart homes are the opposite: cameras send images to the cloud, video doorbells push the ringing video to your phone, vacuum cleaners sync maps and logs to the manufacturer's server, and cloud albums move photos from your phone outward—all of this is sent outward and moving upward.

And this line is asymmetrical. My 300M package only has 30M upstream, and the downstream uplink ratio is 10 to 1. The downstream 300MB is mostly just a few dozen Mbps of downloads at home, so there's plenty of extra space; The 30M uphill lane is the real single-plank bridge where the whole house is crowded. So, when "the network becomes laggy after switching to smart devices," it's almost certainly not because of insufficient broadband, but because this narrow bridge is blocked. This also explains why features like video calls, surveillance playback, and door lock notifications are all stuck — they all rely on uplink support. The uplink bandwidth was stolen, I calculated it once in the post about lagging during a remote video conference. At that time, it was the cloud disk backup, but now it's the cameras' turn.

How can you tell who's eating in 192.168.0.1

? Open 192.168.0.1 in your computer browser and log in to the backend (melogin.cn that domain also runs through the same backend, pick whichever is convenient for you, the pit that can't log in. I wrote about this before

), find 'Device Management.' Some firmware versions have a page called "Host Status" or "Connected Devices," which basically shows a list of online devices: name, MAC, connected frequency band, and the last column shows the real-time rate. The focus was on this last column.

Here are a few practical tips for now. First, the list speed is scrolling. I usually open the page for half a minute to see which devices have numbers that never reset to zero. Second, just looking at the numbers isn't enough. Cooperate with the power cut to identify people: all smart devices at home have physical power outages (camera unplugged, vacuum cleaner recharges and cuts power, doorbell with battery and is paused first). Disconnect one device and check which line in the list drops to zero, matching the numbers one-to-one. This clueless device identification method was written in the article where the device list is full of garbled text. If you can't tell who's who, check that article first. Third, if the rate column occasionally shows negative or exaggerated large numbers, that's due to misaligned page sampling. Don't use it as evidence—run several rounds to get stable values.

My test record from that night, choosing stable numbers, recorded it like this:

when the . video The . The . The , and the . , with events stacked and , of 18M

Indicates	device's daytime stable upward	peak appears
Living room camera	about 1.2M	when people are walking, 2M +	all-weather video uploaded to the cloud
Door camera	about 1.1M	, same as when returning home in the evening	, dual-camera
doorbell	0.4M	moment I pressed the bell, it was 2.2M	; normally I only sent a heartbeat
vacuum cleaner	was less than 0.1M	night I scanned images, it was 2.1M	whole house was being mapped out, with logs spreading like crazy
Two cloud cameras + doorbell totaled	about 3M	at 5M+	Each unit is not large, multiplied by the number of units
My wife's phone cloud album	is silent during the day,	on weekends, charging is 18M	with a maximum single variable

The

most interesting part of this account is the last line. The camera, doorbell, and sweeper add up to around 3M in daily life, which doesn't look scary; The real big seller is my wife's cloud album on her phone. Once you start charging, it automatically backs up, and you can single-million-dollar it can consume 18MB. If it takes 18M of 30M uplink speed, my mom's video calls will get blurred. Also, that night I re-scanned the entire house with the vacuum cleaner, and the log sync stacked up another 2M—several 'not big' ones collided in the same night, and the upstream was completely exhausted. This is the same idea as the idea of 'using router tools for home with slow internet speeds in three steps to find and grab bandwidth devices.' The specific usage on the tool page is written in more detail here.

Why does speed testing software trick you

?

Here, I need to insert a segment, because I was misled by the speed testing software from the start. During that time, I was running speedtests in the living room, 320M downhill and 28M uphill speed. Everything looked normal, so I confidently passed the blame to broadband. Later, I realized two things.

First, speed measurement is a snapshot. The cloud album only started backing up after charging on weekends. I chose a weekday evening for speed testing, but I missed it. To find out the truth, you have to check during the "stuck time period"—my mom said the video blurry is mostly when cloud backup is running. Second, speed testing itself is also a rush. The speed testing software fed data to a remote server to test uplink, and it itself had the largest uplink traffic at the time, occupying dozens of Mbps out of 30M, while other devices were squeezed out by it. So if you have smart devices at home, you should discount your speed test results. Don't treat them as medical reports—just use them as references. The truly reliable method is the backend rate column plus power-off method—slow or slow, but the numbers are real.

Three ways to fix it: two routers that don't work, one that just started

checking,

and then fixing it. I follow the order of "device side first, router last," which is important: the device side is the source, solving it once and for all; Router speed limiting is like cutting the pie by force; if you don't cut it well, you'll hurt yourself. Let's go over three reform methods one by one.

First, move cloud backup to midnight and limit uplink to 10M. In the backup settings for my wife's phone cloud album, I set "Backup only while charging" and set the start time to 1 a.m. This frees up 18M whales from the prime daytime hours. Just the time spent moving wasn't enough—if I plugged my phone into a power bank during the day, the backup would still wake up, so I set a 10M uplink limit on the router, and even if it woke up, it wouldn't be fully charged. Cloud backup shifting time slots is the same principle as NAS users staggering tasks. Specifically, on routers, how to limit the speed of a single device: Mercury opens the device in device management and it has uplink and downlink options. Tenda's new firmware is also set by device in device management. I < the entry points of both companies a href="https://www.shunot.com/en/192-168-0-1/824.html"> Mercury speed limits written about Tenda Speed Limit in two separate articles.

Second, the camera switched to event recording, and the doorbell originally only carried the event. The two cameras were originally designed for round-the-clock cloud recording, but after switching to motion detection, they only send heartbeats when no one is around, and during the day, the uphill speed drops from just over 3M to less than 1M; The resolution was easily reduced from 1080P to 720P, with no significant difference in playback, and bitrate was cut in half. All these changes are made in the manufacturer's app without touching the router, making it the most cost-effective among the three modification methods. A reminder: After changing to event recording, you can't review footage during "no one" period. If you have pets at home or need to record all day, skip this point—don't sacrifice essential features just to save bandwidth.

Only the third one is the router to leave a backup for the WAN port to uplink the total uplink. My approach isn't to lock down any device, but to do the opposite: after suppressing the uplink of several devices that will load (the phone in the cloud album, the robot vacuum), I ensure there is always some uplink during video call sessions. There's a pitfall here that must be mentioned—the unit of the speed limit box. The old Tenda model had a Kbps frame; if you want to limit 10MB, you have to enter 10240; The new Mercury model is marked with Mbps, so just enter 10. If the workplace misreads, it either restricts the device to a few KB per second, making WeChat voice messages impossible. The unit conversion account is covered in the article Speed Limit Five-Step Guide, which covers the whole section. Go through it before starting the router to avoid unnecessary detours.

Which symptoms are related to the upward movement, and which really have nothing to do

with it?

After making these two weeks, I categorized all the symptoms that had occurred at home by whether they were the 'upward cause' to avoid misdiagnosis next

time Is an ? ? During . . . . The

the symptom	upward pot	What should be done first
video calls, the other party is mostly seeing you as a foul/card	(your uplink isn't enough to push it outward),	checking who is uploading, cloud backup is the top suspect
Surveillance replays spinning and door lock notifications arrive late	. (Crowding the channel for outbound broadcasts)	first edit the event footage, then consider speed limiting
Watch the 4K live streaming card	No, that's not (this is about downlink + WiFi).	Check the box's speed limit rules and channels
High game latency and skills not	being released may not be necessary (low data, check packet loss and WiFi quality).	First, check signal and wired wireless
webpage takes half a day to	open very little (DNS is more suspicious).	Change DNS before speaking

The biggest use of this watch is to block the urge to switch broadband with just one card. Downlink expansion can't solve upstream congestion; 300M for 1000M, uplink still just a few dozen meters, and the cameras are still the same. If you really want to expand, first look at how much the uplink of the plan offers—that's the bottleneck for smart device families. By the way, speed limiting has another way to play in shared rental scenarios—how to negotiate and divide between peers. I wrote about it in the article splitting broadband in shared rentals, which are two completely different issues from the smart home article.

Wrap-up: After leaving 5M uplink, what's the state of my home

?

I'll give you a bottom line on the current steady state: Cloud album runs automatically at midnight, limited to 10M; Camera event recording adds 720P, totaling less than 1MB during the day; The whole-house sweep was manually scheduled late at night; During video call sessions, the uplink margin is always over 5M. My mom's video calls never blurred again; as soon as the doorbell rang and her phone rang, surveillance replays came at any moment. The final actions on the router side were actually very limited—identifying the device list, limiting it to two, checking the WAN port speed settings, and less than an hour in total, with most of the time spent waiting in the background and cutting power off one by one.

If your home also reaches the stage of "ten smart devices or more, calls start to get blurry, and surveillance starts spinning," my advice is three points: first open the backend to check the speed list and identify people, don't rush to order gigabit broadband; The main issue is first handled on the device side, with cloud backup for shifting time slots and camera event changes, which is more effective than speed limiting; The router's speed limit is only a backup; before starting, make sure to check both the unit and direction. When I found it hard to recognize which device was which, I went back < the article about a href="https://www.shunot.com/en/192-168-0-1/1540.html" > device identification and started with the code code.