Testing Bandwidth Vs. Low Light

By Ethan Ace, Published Jan 16, 2019, 03:34pm EST

Nighttime bandwidth spikes are a major concern in video surveillance. Many calculate bandwidth as a single 24/7 number, but bit rates vary dramatically from day to night. But how much? 100% more? 500%? 1,000%?

day night

To find out we tested cameras from six manufacturers - Avigilon, Axis, Dahua, Hanwha, Hikvision, Uniview - to find out how each did, what was average, and who was worst and best and how much performance has improved in the past several years.

Key ********

***** *** * *** key ********* **** **** test:

  • ~***% ******* ********:****** ** ******* ******* from ***** ** **, bit **** ********* ** average ** ~***% **** day ** *****. ********* varied **** * *** of ~**% ** **** 400%.
  • ******* ******* ****** ***** past *****:*** ***% ******* **** in **** **** ** roughly **% ***** **** the ***% ******* ** our **** *****.
  • ** ******** ******:*** ******* ****** ** performance *********** ** **** vs. ******* ******* *** ********* ** integrated **, *** ***** in *** ******** ** cameras, ***** ** *** very ******* ******* ***** ago.
  • *** (*** **** * cap) ***********:** ****** ********* ******, we ********* ***** ******* bit **** ****, ** this ****** *** ***** to **** ** *** as ******** ****** **** complex *****, *** **** bandwidth ** * ********* limit ** ***** ****** network ******* *** *******, with ******* ****** ** video *******.

**** **** ***** **** no ********** ****** ** how ****** ****** ***** on ************ ** **********. Each ************ *** ********** varied ******, *** ****** sizes *** ***** ** make * **** *************.

145% ******** *******

****** ***** ** *******, the ******* (****) ********* increase **** *** ** night *** * ***%. Note **** *** ****** was "****" ~**%, **** the ***% ******* ****** higher *** ** ******** 400%+ ********* ** **** cameras.

**** *** *** ******* arranged ** **********:

2019

Much ******* ******** **** **** *****

** ******** ** *** 145% ******** **** ** current *****, ******* ** our **** **** ******** 498% ****** ********* ** night, **** *** ****** spiking ****** *,***% ******.

The **********: ********** ** *** **** *********

*** **** ********* ****** change ** ****** *********** between *** ******** **** in **** *** ****** tested ** **** ** the ********* ** ********** IR ** ****** ***** camera ******, ******** ** only * ****** ** 2014 **** *** ******* was ***** ****.

********** ****, ** ** look ** **** ***** integrated ** ****** **** 2014, *********** *** **** more ******* ** **** averages, * ***% ******* increase.

2014 only

***** ****** *********** *** improved, **** **** ********* encoders, ****** *** **** sensitive *******, *** ****** processing, *** ******** ** infrared ***** ** ********* surveillance *** *** ***** positive ****** ** ********.

Root ***** ** ********* ******: ******** *****

*** ****** ** ***** spikes ** *** ******* moving *****, *** ****, static, *******, ** **** that ** ********* *********** **** *******.

** ***** **** **:

** ** *******, **** noise, **** *** **** contrast (***** *** ***** near **** *****) *** rapid ********* (******* ******), make ** ****** ** encode. ******,********* ******* *** ****** / ***** **** *** similar *** **********. ***** undermines **** ** *****.

Recommendation: *** *** **** * ***

*** *** ** ***** massive ********* ****** *** be ******* ** ******* *** **** * cap, **** ***** ** MBR(******* *** ****). **** method ****** ******** ** fluctuate ** ********* ** fit *** *****, *** only ** ** * specified *****, ** ***** point *** ****** ********* compression ** ****** ** drop ****** ** **** within *** *** (********* the ****** ** *********).

** *** *****, **** when **** ******************** ********* **** ** average ** ~** ** 35, ***** ******* *********** were *******, **** ****** bandwidth ***** ** **** 80%

bandwidth cap minimal impact on details

Constant *** **** *** ***********

******** *** **** *** also ** **** ** prevent ******, ***** ** locks *** ***** ** one ********* ***** ********** of ******** ** *** scene. *******, ****** *** day, **** ****** *** less *******, *** *** waste *********/******* ** *** camera ***** **** *********** encode *** ***** **** set ** ***. ******* of ****, *** ** not *********** ****** ** the **** ****** ***** does *** ******* ***.

Digital ***** ********* ******

******* ****** ********* ***** is **** ** ****** spikes *** ** ********* noise ** ******* ***** reduction (***), ***** ******** to ******* ***** ****** by ********* ** ****. There *** *** *** methods, ******* (***** ********** in * ****** *****) and ******** (***** ********* by *********** ******** ******), which *** ********* ******** in ******. ***** ***, spikes *** ** *********** cut, **** **-**% ********** in *** **** ******** *** *****.

*******, *** *** **** introduce ******** *******, ** well, **** **** ****** of ******** ***** ********* causing **** ** ***** (shown *****), ********* ******* to *** ******* ** slow *******. ******* ** this, ***** ****** ** careful ** *** ******** DNR ** ******* ******.

*** **** ******* *** testing, *** *** ****** *** (******* ***** Reduction) ***** *** ****.

Test *******

** ****** ** ******* from *** ************* *** (~300 ***) *** ***** (~0.1 ***) ******** ****** scenes. *** ******* *********** levels **************** ***-** ****** ******* smart ****** **. ***** codecs **** ******* ** their ******* ******.

[****: **** **** *** originally ******** ** ****. We ********** ***** *** test ** **** *** republished *** ****** **** the *** ******* *** contrasting *** *** ****.]

Comments (8)

The numbers in the "Day" and "Night" columns are Mbps, correct?

Yessir.

I have found that turning the sharpness setting on the camera down a bit helps with the high nighttime bandwidth.

We've had quite a bit of experience with this since the nighttime bandwidth overran our wireless network capacity at night.  We use Hikvision cameras and made the following settings:

- WDR set to 25 (range is 0..100, default is 50) - decreases dynamic range sensitivity

- DNR set to 75 (range is 0..100, default is 50) - increases noise filtering

- Sharpness set to 25 (range is 0..100, default is 50) - decreases sharpness

 

All of these settings decrease the quality of the video especially during the day.  The sharpness setting muted the vibrancy of the daytime video signal and in many cases on a camera by camera basis we had to back this value back off to a value closer to the default value.

The 4K cameras were the worst offenders for nighttime bandwidth blooming.  In some cases, the camera ignored the MBR cap we set which was disappointing.  Only just a few cameras had the blooming problems but we saw daytime bandwidth increase from 4 mbps during the daytime to 20 mbps at night and occasionally as much as 100 mbps as reported by the Milestone Smart Client System Monitor.  The bandwidth increases were not constant and varied considerably between samples in the System Monitor.  We were using the H.264+ CODEC on all the cameras.

Are all of these mpbs values with H264+ codecs? As a comparison, it would also be interesting to see if H265 makes a difference in this scenario.

Yes, this is all H.264. We agree that H.265 would be an interesting addition here, but even in 2019 it significantly reduces the number of cameras we'd include in the test. As Avigilon, Axis, Bosch, and others get more H.265 models released, we'll look at this again.

Axis seen to de very well here, can you attribute this to anything in particular? 

I think it's a combination of factors. Their cameras are generally among the lowest bitrate models we test in most categories. It's pretty rare in 2018/2019 that they don't come in competitively low in our tests, day or night.

They are using their own chip design in the majority of their cameras (ARTPEC), where many competitors are using HiSilicon or Ambarella. Some of their cameras actually use both the ARTPEC chip plus another SoC to split processing. So, I believe that since they're working with their own chip, it could simply be a matter of better "tuning" for more efficient compression.

Additionally, Axis cameras tend to (usually, but not always) be good in low light. They are often, again, among top performers for low light imaging and their IR coverage is generally more even than many competitive cameras.

When you combine all of it, I think it all helps to cut down spikes.

This is not to say that Axis is hands down best. A number of Hikvision models have been performing well, too, especially in more recent Performance and G0/G1 series cameras. And newer Hanwha models tend to not spike nearly as much as older Hanwha/Samsung cameras, which were notorious.

Read this IPVM report for free.

This article is part of IPVM's 6,803 reports, 913 tests and is only available to members. To get a one-time preview of our work, enter your work email to access the full article.

Already a member? Login here | Join now
Loading Related Reports