Video quality under degraded network conditions is one of the most important factors when evaluating a real-time communication SDK.
Under ideal network conditions, many video SDKs can deliver similar results. The differences become much clearer when users experience packet loss, bandwidth constraints, or high jitter — conditions commonly encountered on mobile networks, congested Wi-Fi, elevators, basements, and cross-border connections.
In this benchmark, we compared Tencent RTC (TRTC) and Vonage Video API under controlled weak-network conditions.
Both SDKs used the same video source, comparable configurations, the same mobile devices, and the same professional network impairment environment.
Key Results at a Glance
Test configuration:
Sender: iPhone 12 Pro
Receiver: iPhone 14
Video: 640×480
Target frame rate: 30 fps
Target bitrate: 800 Kbps
Weak-network scenarios: 8
Network impairments: packet loss, bandwidth limitation, and jitter
| Metric | TRTC | Vonage Video API |
|---|---|---|
| Average frame rate across weak-network scenarios | 28.2 fps | 14.7 fps |
| Resolution retention at 640×480 | 100% | 48% |
| Resolution retention during recovery window | 100% | 16% |
| Average receiver CPU usage | 15.1% | 20.6% |
| Average receiver memory usage | 62 MB | 55 MB |
Across the eight weak-network scenarios, TRTC delivered an average frame rate of 28.2 fps, compared with 14.7 fps for Vonage, representing approximately 91% higher average frame rate in this test.
TRTC also remained at the configured 640×480 resolution for 100% of the weak-network test periods, while Vonage remained at 640×480 for an average of 48% of the time.
After network conditions returned to normal, TRTC remained at 640×480 throughout the recovery observation window. Vonage operated at 640×480 for an average of 16% of that recovery window.
Test Methodology
Both SDKs were tested under identical controlled conditions.
Frame-rate measurements refer to frames actually rendered on the receiving device, rather than frame-rate values reported internally by the SDK.
Mobile Test Setup
| Item | Configuration |
|---|---|
| Sender | iPhone 12 Pro |
| Receiver | iPhone 14 |
| TRTC SDK | 13.5.0.21355 |
| Vonage SDK | 2.34.1 |
| Video resolution | 640×480 |
| Target frame rate | 30 fps |
| Target bitrate | 800 Kbps |
| Network impairment | Professional hardware network impairment device |
Test Scenarios
Each scenario followed the same general sequence:
Establish baseline → apply network impairment → remove impairment → observe recovery
| Scenario | Network Condition | Duration |
|---|---|---|
| Normal Network | Normal network conditions used as baseline | 60s |
| 25% Packet Loss | Uplink and downlink tested separately | 30s each |
| 500 Kbps Bandwidth Cap | Uplink first, followed by downlink | 60s |
| 500 Kbps + 25% Packet Loss | Uplink first, followed by downlink | 60s |
| 600ms Jitter | Uplink and downlink tested separately | 30s each |
These conditions are intended to simulate situations such as: mobile users connected through congested cellular networks; video calls in elevators or basements; unstable Wi-Fi connections; constrained ISP bandwidth; cross-region or cross-border connections with significant network variation.
Test constraints
This report presents TRTC performance with the “Ensuring video smoothness” preference, i.e. the setting that prioritizes video smoothness over image clarity. This is an account-level setting that is disabled by default; to reproduce the test, contact TRTC Technical Support to have this setting provisioned for your account.
Mobile Video Quality Results
Frame Rate Under Weak-Network Conditions
Across all 8 weak-network scenarios:
TRTC average: 28.2 fps
Vonage average: 14.7 fps
The complete results are shown below.
| Weak-Network Scenario | TRTC | Vonage | TRTC Difference |
|---|---|---|---|
| Uplink 25% Packet Loss | 29.3 fps | 16.6 fps | +77% |
| Downlink 25% Packet Loss | 30.1 fps | 18.4 fps | +64% |
| Uplink 500 Kbps Cap | 31.2 fps | 15.7 fps | +99% |
| Downlink 500 Kbps Cap | 26.8 fps | 20.4 fps | +31% |
| Uplink 500 Kbps + 25% Loss | 21.4 fps | 10.1 fps | +112% |
| Downlink 500 Kbps + 25% Loss | 26.8 fps | 24.1 fps | +11% |
| Uplink 600ms Jitter | 29.3 fps | 7.0 fps | +319% |
| Downlink 600ms Jitter | 30.4 fps | 5.3 fps | +474% |
| Average | 28.2 fps | 14.7 fps | +91% |
Key Observation
- In 5 of the 8 weak-network scenarios, TRTC maintained frame rates above 26 fps while remaining at 640×480 resolution throughout the test period.
- Across the other three scenarios, TRTC's average frame rate remained between approximately 21 - 27 fps.
- Vonage showed greater frame-rate variation across the same scenarios, ranging from 5.3 fps to 24.1 fps.
- Vonage also reduced the received resolution below 640×480 in 7 of the 8 weak-network scenarios.
Frame Rate Over Time Under Weak Networks
The average frame rates above summarize overall performance, but they do not show how video quality changes at the moment network impairment begins or how quickly performance stabilizes after the network returns to normal.
The chart below shows representative frame-rate curves across the baseline, impaired-network, and recovery stages.
- Y-axis: receiver frame rate (fps)
- X-axis: time within scenario
- Orange-shaded region: uplink restrictions active
- Red-shaded region: downlink restrictions active
- Green-shaded region: recovery observation window
Figure: Representative TRTC and Vonage Video API frame-rate changes over time.
Scenario-by-Scenario Results
25% Uplink Packet Loss
A 25% packet-loss condition was applied to the sender's uplink.
| Metric | TRTC | Vonage |
|---|---|---|
| Average frame rate | 29.3 fps | 16.6 fps |
| Time at 640×480 | 100% | 40% |
| Lower-resolution usage | 0% | 60% |
TRTC remained at 640×480 throughout the test while averaging 29.3 fps.
Vonage averaged 16.6 fps and remained at 640×480 for approximately 40% of the impaired period. During the remaining time, the received video operated at lower resolutions such as 480×360 or 320×240.
25% Downlink Packet Loss
A 25% packet-loss condition was applied to the receiver's downlink.
| Metric | TRTC | Vonage |
|---|---|---|
| Average frame rate | 30.1 fps | 18.4 fps |
| Time at 640×480 | 100% | 30% |
| Lower-resolution usage | 0% | 70% |
TRTC maintained approximately 30 fps while remaining at the configured 640×480 resolution.
Vonage averaged 18.4 fps and remained at 640×480 for approximately 30% of the test period.
500 Kbps Uplink Bandwidth Cap
The sender's available uplink bandwidth was limited to 500 Kbps.
| Metric | TRTC | Vonage |
|---|---|---|
| Average frame rate | 31.2 fps | 15.7 fps |
| Time at 640×480 | 100% | 45% |
Under the 500 Kbps uplink constraint, TRTC remained at 640×480 throughout the test and averaged 31.2 fps.
Vonage averaged 15.7 fps and remained at 640×480 for approximately 45% of the constrained period.
500 Kbps Downlink Bandwidth Cap
The receiver's downlink bandwidth was limited to 500 Kbps.
| Metric | TRTC | Vonage |
|---|---|---|
| Average frame rate | 26.8 fps | 20.4 fps |
| Time at 640×480 | 100% | 100% |
This was the only weak-network scenario in which both SDKs remained at 640×480 for the entire constrained period.
TRTC averaged 26.8 fps while Vonage averaged 20.4 fps.
500 Kbps + 25% Uplink Packet Loss
A 500 Kbps bandwidth restriction and 25% packet loss were applied simultaneously to the sender's uplink.
| Metric | TRTC | Vonage |
|---|---|---|
| Average frame rate | 21.4 fps | 10.1 fps |
| Time at 640×480 | 100% | 35% |
This was the most demanding uplink condition in the benchmark.
TRTC averaged 21.4 fps while maintaining 640×480 resolution throughout the impaired period.
Vonage averaged 10.1 fps and remained at 640×480 for approximately 35% of the period.
500 Kbps + 25% Downlink Packet Loss
A 500 Kbps bandwidth cap and 25% packet loss were applied simultaneously to the receiver's downlink.
| Metric | TRTC | Vonage |
|---|---|---|
| Average frame rate | 26.8 fps | 24.1 fps |
| Time at 640×480 | 100% | 25% |
The difference in average frame rate was relatively small in this scenario.
However, the received resolution behaved differently: TRTC remained at 640×480 for the entire period, while Vonage remained at 640×480 for approximately 25% of the period.
600ms Uplink Jitter
600ms jitter was introduced on the sender's uplink.
| Metric | TRTC | Vonage |
|---|---|---|
| Average frame rate | 29.3 fps | 7.0 fps |
| Time at 640×480 | 100% | 40% |
TRTC averaged 29.3 fps while remaining at 640×480 throughout the test.
Vonage averaged 7.0 fps and remained at the configured resolution for approximately 40% of the impaired period.
600ms Downlink Jitter
600ms jitter was introduced on the receiver's downlink.
| Metric | TRTC | Vonage |
|---|---|---|
| Average frame rate | 30.4 fps | 5.3 fps |
| Time at 640×480 | 100% | 70% |
This scenario produced the largest frame-rate difference in the benchmark.
TRTC averaged 30.4 fps while Vonage averaged 5.3 fps.
TRTC remained at 640×480 throughout the impaired period, while Vonage remained at that resolution for approximately 70% of the time.
Resolution Retention Under Weak Networks
Frame rate is only one part of perceived video quality.
Another important question is:
How often does the received video remain at the configured 640×480 resolution when the network deteriorates?
The following table shows the percentage of each weak-network period spent at 640×480.
| Scenario | TRTC at 640×480 | Vonage at 640×480 |
|---|---|---|
| Uplink 25% Packet Loss | 100% | 40% |
| Downlink 25% Packet Loss | 100% | 30% |
| Uplink 500 Kbps Cap | 100% | 45% |
| Downlink 500 Kbps Cap | 100% | 100% |
| Uplink 500 Kbps + 25% Loss | 100% | 35% |
| Downlink 500 Kbps + 25% Loss | 100% | 25% |
| Uplink 600ms Jitter | 100% | 40% |
| Downlink 600ms Jitter | 100% | 70% |
| Average | 100% | 48% |
Across these eight scenarios, TRTC remained at 640×480 for the entire weak-network period.
Vonage remained at 640×480 for an average of 48% of the test period, with the remaining time spent at lower resolutions such as 480×360 or 320×240.
Why Resolution Retention Matters
Resolution loss is particularly important in applications where users need to see fine visual details.
| Use Case | Why Resolution Matters |
|---|---|
| Video KYC | Facial details and identity documents must remain readable |
| Live commerce | Product texture, appearance, and other visual details affect product presentation |
| Remote consultation | Small visual details may be important during a consultation |
| Online education | Whiteboard text, documents, and demonstrations need to remain readable |
| Video customer service | Agents may need to identify products, devices, or documents shown by customers |
At lower resolutions such as 320×240, fine visual information becomes significantly harder to distinguish.
Recovery After the Network Returns to Normal
Weak-network handling is only part of the experience.
Another important question is how quickly video quality returns after network conditions improve.
In these tests, frame-rate recovery was relatively fast for both SDKs.
Resolution recovery showed a larger difference.
| Recovery Metric | TRTC | Vonage |
|---|---|---|
| Average time at 640×480 during recovery observation window | 100% | 16% |
The recovery observation period was 30 seconds.
TRTC remained at 640×480 throughout the recovery window.
In this test, Vonage remained at a reduced resolution during much of the recovery period even after the network impairment had been removed.
From an end-user perspective, an unresolved network problem and delayed SDK resolution recovery can appear similar: both result in a video image that remains blurry.
Frame-Rate Recovery Time
| Scenario | TRTC Recovery | Vonage Recovery |
|---|---|---|
| Uplink 25% Packet Loss | 0 s | 0 s |
| Downlink 25% Packet Loss | 0 s | 1.0 s |
| Uplink 500 Kbps Cap | 0 s | 0 s |
| Downlink 500 Kbps Cap | 0 s | 7.0 s |
| Uplink 500 Kbps + 25% Loss | 0 s | 0 s |
| Downlink 500 Kbps + 25% Loss | 0 s | 0 s |
| Uplink 600ms Jitter | 0 s | 4.0 s |
| Downlink 600ms Jitter | 0 s | 3.0 s |
It is important to distinguish frame-rate recovery from resolution recovery.
Frame rate could recover quickly while resolution remained below the configured 640×480 level.
CPU and Memory Usage
Resource consumption is also important for mobile RTC applications, particularly during long calls or streams.
Measurements were collected on the receiving device:
Device: iPhone 14
Chip: A15 Bionic
CPU measurement: 100% represents one fully utilized CPU core
Average Resource Usage
| Metric | TRTC | Vonage |
|---|---|---|
| Average CPU usage | 15.1% | 20.6% |
| Average memory usage | 62 MB | 55 MB |
TRTC used approximately 5.5 percentage points less CPU on average across the tested scenarios.
Vonage used less memory on average, at approximately 55 MB compared with 62 MB for TRTC.
CPU and Memory by Scenario
| Scenario | TRTC CPU | Vonage CPU | TRTC Memory | Vonage Memory |
|---|---|---|---|---|
| Normal | 13.9% | 24.2% | 60 MB | 51 MB |
| Uplink 25% Packet Loss | 15.2% | 19.5% | 60 MB | 64 MB |
| Downlink 25% Packet Loss | 15.2% | 20.0% | 60 MB | 49 MB |
| Uplink 500 Kbps | 14.9% | 23.5% | 74 MB | 64 MB |
| Downlink 500 Kbps | 14.5% | 20.4% | 68 MB | 46 MB |
| Uplink 500 Kbps + 25% Loss | 14.7% | 22.7% | 60 MB | 65 MB |
| Downlink 500 Kbps + 25% Loss | 15.2% | 19.3% | 60 MB | 45 MB |
| Uplink 600ms Jitter | 15.7% | 18.9% | 61 MB | 57 MB |
| Downlink 600ms Jitter | 16.4% | 16.6% | 61 MB | 53 MB |
| Average | 15.1% | 20.6% | 62 MB | 55 MB |
CPU usage can be particularly relevant for long-running video sessions because sustained CPU load may contribute to battery consumption and thermal pressure on mobile devices.
What These Results Mean in Practice
Different applications prioritize different aspects of video quality.
Some scenarios depend heavily on visual detail, while others prioritize smooth motion or lower device resource consumption.
Interactive Live Streaming
| Scenario | Critical Metric | Potential User Impact |
|---|---|---|
| E-commerce live streaming | Resolution retention | Product textures, visual details, and appearance are easier to inspect at higher resolution |
| Outdoor or event streaming | Frame rate under impairment | A weak host uplink can affect the video experience seen by viewers |
| Multi-host or guest interaction | Frame rate | Large differences in motion smoothness between participants can become visible |
| Long-duration streaming | CPU usage | Lower CPU usage may reduce thermal pressure during extended sessions |
For live-commerce applications in particular, resolution retention may be important because customers often rely on video to inspect product details.
1:1 Video Calls
| Scenario | Critical Metric | Potential User Impact |
|---|---|---|
| Video KYC / remote account opening | Resolution retention | Face and document details need to remain readable |
| Social video / dating | Resolution + frame rate | Both image detail and motion smoothness contribute to perceived quality |
| Video interviews | Frame rate under jitter | Participants may connect from unstable home, mobile, or cross-region networks |
| Lower-end device support | CPU usage | Lower resource usage may reduce thermal and performance pressure |
Conversational AI and Video Agents
For AI-powered voice and video agents, stable video input can also influence downstream processing.
Weak-network conditions may introduce: frame drops; resolution changes; inconsistent visual input; unstable facial or object details.
These changes may affect applications that use incoming video for tasks such as: facial analysis; visual context understanding; gesture recognition; emotion-related analysis; real-time multimodal AI interaction.
The frame-rate and resolution-retention results in this benchmark may therefore also be relevant when evaluating RTC infrastructure for AI-driven video applications.
Different SDK Behaviors Under Bandwidth Constraints
The measured results suggest that the two SDKs behaved differently under constrained network conditions.
These observations are based on output measurements from the benchmark and should not be interpreted as confirmed information about either SDK's internal algorithms or implementation.
TRTC Behavior Observed in This Test
Across all eight weak-network scenarios, TRTC remained at 640×480 for 100% of the impaired period.
Frame rate varied according to the severity of the network condition.
The lowest average frame rate measured for TRTC in the eight scenarios was:
21.4 fps under a 500 Kbps uplink cap combined with 25% packet loss.
This suggests that, under the tested configuration, TRTC generally retained the configured resolution while allowing frame rate to vary as network conditions changed.
Vonage Behavior Observed in This Test
Vonage reduced the received resolution below 640×480 during 7 of the 8 weak-network scenarios.
Across all eight scenarios, it remained at 640×480 for an average of 48% of the impaired period.
| Scenario | Vonage Frame Rate | Time at 640×480 |
|---|---|---|
| Uplink 25% Packet Loss | 16.6 fps | 40% |
| Downlink 25% Packet Loss | 18.4 fps | 30% |
| Uplink 500 Kbps Cap | 15.7 fps | 45% |
| Downlink 500 Kbps Cap | 20.4 fps | 100% |
| Uplink 500 Kbps + 25% Loss | 10.1 fps | 35% |
| Downlink 500 Kbps + 25% Loss | 24.1 fps | 25% |
| Uplink 600ms Jitter | 7.0 fps | 40% |
| Downlink 600ms Jitter | 5.3 fps | 70% |
Which adaptation strategy is preferable depends on the application.
Applications where visual detail is important — such as identity verification or product presentation — may place greater value on maintaining resolution.
Applications where continuous motion is particularly important may assign greater weight to frame-rate behavior.
Complete Mobile Benchmark Data
For easier comparison and machine-readable reference, the primary benchmark results are consolidated below.
| Scenario | TRTC FPS | Vonage FPS | TRTC Resolution Retention | Vonage Resolution Retention |
|---|---|---|---|---|
| Uplink 25% Packet Loss | 29.3 | 16.6 | 100% | 40% |
| Downlink 25% Packet Loss | 30.1 | 18.4 | 100% | 30% |
| Uplink 500 Kbps Cap | 31.2 | 15.7 | 100% | 45% |
| Downlink 500 Kbps Cap | 26.8 | 20.4 | 100% | 100% |
| Uplink 500 Kbps + 25% Loss | 21.4 | 10.1 | 100% | 35% |
| Downlink 500 Kbps + 25% Loss | 26.8 | 24.1 | 100% | 25% |
| Uplink 600ms Jitter | 29.3 | 7.0 | 100% | 40% |
| Downlink 600ms Jitter | 30.4 | 5.3 | 100% | 70% |
| Average | 28.2 | 14.7 | 100% | 48% |
Benchmark Summary
Under the test configuration used in September 2026, TRTC and Vonage Video API showed measurable differences in weak-network behavior.
The main findings were:
TRTC averaged 28.2 fps, compared with 14.7 fps for Vonage, across the eight weak-network scenarios.
TRTC remained at 640×480 for 100% of the weak-network test periods, compared with 48% for Vonage.
During the recovery observation window, TRTC remained at 640×480 for 100% of the time, compared with 16% for Vonage.
TRTC averaged 15.1% CPU usage, compared with 20.6% for Vonage.
Vonage used less memory on average: 55 MB compared with 62 MB for TRTC.
The largest frame-rate differences appeared in the 600ms jitter tests.
The smallest frame-rate difference appeared under 500 Kbps + 25% downlink packet loss, where TRTC averaged 26.8 fps and Vonage averaged 24.1 fps.
These results apply specifically to the devices, SDK versions, video configuration, and network conditions described in this benchmark.
Performance may vary with other devices, SDK releases, encoding settings, network routes, or application configurations.
Test Information
Test date: September 2026
TRTC SDK: 13.5.0.21355
Vonage SDK: 2.34.1
Sender: iPhone 12 Pro
Receiver: iPhone 14
Video configuration: 640×480, 30 fps, 800 Kbps target
All benchmark data reflects the SDK versions and test configurations described in this article.
Results may vary under different network conditions, device combinations, SDK versions, and application configurations.
For questions about this benchmark or Tencent RTC, visit Tencent RTC Contact.
