Matrix: when do I need to change the LAN setting or the jitter buffer settings on your product to improve connectivity issues.
'We have had a number of customers that have come to us asking when do they need to change the LAN setting or the jitter buffer settings on your product to improve connectivity issues'.
There are far too many variables when going over IP networks to allow us what state what the best setup would be depending on the network topology. With regard to AoIP products the decision is in the ‘ear of the beholder’, and by that we mean if the users are experiencing choppy audio (i.e. packet loss). Packet loss can occur anywhere within the network and therefore, several countermeasures can be applied to limit the amount of packet loss depending on the root cause:
A)Jitter: Jitter is a behavior where the speed at which packets are received over the network varies due to a variety of possible QoS issues. Without a jitter buffer,any packet received later than expected would be dropped by the target device. The user would observe the dropped packet as silence but due to the high frequency of packet transmission, a few dropped packets would sound like choppy audio as the dropped packets would in essence create holes within the audio stream.
To address this, Jitter buffers have been implemented to buffer the received packets for a certain duration before the audio is played back to the user. This buffering prevents dropping of packets that are received later than expected or even packets received out of order. The result is a cleaner sounding audio experience but at a cost. The minimum buffer size selected has a direct impact on latency.
The lowest selectable jitter buffer setting is 5ms – 60ms.
If a 20ms – 60ms jitter buffer is selected, then latency observed by the listener will be increased by at least 15ms. One more thing to note is that the jitter buffer implementation in Clear-Com products is dynamic. This means that if jitter is observed by the unit to be higher than the minimum buffer size (20ms in this example), then the buffer will dynamically increase or decrease to account for the variance but only up to the maximum size specified (60ms in this example).
Therefore, with this 20ms – 60ms setting, if a packet is received more than 60ms late, then that packet will be dropped and choppy audio will be observed until the buffer is once again increased to handle the excess jitter. Depending on the product, the jitter buffer can be modified either manually or by decreasing the overall network quality setting.
B)Audio Bitrate : The Audio Bitrate setting defines the level of information captured for each audio sample. The higher the setting the bigger the packet will be resulting in a higher network bandwidth utilization. If a high Audio Bitrate is selected while utilizing a low bandwidth network, packet loss would be the likely outcome once again resulting in the observance of choppy audio. Therefore, it is recommended to decrease the Audio Bitrate when operating within a low bandwidth environment.
Depending on the product, the Audio Bitrate can be modified either manually or by decreasing the overall network quality setting.
C)Packet Size: PacketSize, or sometimes referred to as Frame Duration, defines the amount of audio information, measured in milliseconds, that each packet contains. So a Packet Size setting of 5ms will result in each packets being transmitted to contain 5ms worth of audio. It is important to also understand that packet size and network bandwidth have an inverse relationship.
The reason for this is that each packet is constructed with both a header and a payload. The packet header is always the same size and contains routing and other information to ensure proper packet delivery. The payload portion of the packet contains the audio data. Reducing the Packet size just reduced the payload and not the header which ultimately results in a higher overhead per packet transmitted as. Sending a payload that is 10x larger than the header is much more efficient that sending a payload that is only 5xlarger.
Depending on the product, the packet size can be modified either manually or by decreasing the overall network quality setting.
D)Forward Error Correction (FEC) In certain products, FEC has been implemented to provide a method for packet reconstruction on the receiving end. The concept is that each packet transmitted contains information about previous packets which can be used to reconstruct a lost packet when required.
Normally, enabling this feature will increase the amount of information contained within each packet but with some network quality settings, the bitrate is lowered to counter the increased data ultimately resulting in a wash.
Note that in order to fill packets ready for transmit with prior packet information, the transmit mechanism must wait for the duration of the previous packet’s duration before sending. This ultimately results in an increased latency originating at the sender by a factor equal to the amount of the FEC level specified.
If the Network Quality adjustments stated above resolve the issues being experienced but with too great a cost on audio quality and/or latency, then addressing limitations of the Network infrastructure will need to be addressed with your IT professional or Internet Service Provider (ISP).
Clear-Com can only provide the tools to counter poor network quality environments which our equipment may be operating within. That said, effort has still been invested in making these tools simple to use as possible through the abstraction of labelled configuration sets which attempt to coalesce the best combination of settings for a given environment which can then be selected within the user interface.
Clear-Com recommends the use of the highest network quality setting possible so long as clean audio is observed. Therefore, begin testing with the highest network quality setting and only lower it if anomalies are observed.
Note that the law of the lowest common denominator directly applies to network transport and as a result, any limitation between two endpoints will directly affect the transmission as a whole including bandwidth limitations.
For example, having a 1 Gb/s link on your local LAN segment but only a 10Mb/s link between routers or gateway devices will result in an overall throughput of no greater than 10Mb/s.
. LAN -> Local
*Optimized for lowest latency in stable, controlled networks. Minimal buffering, no packet recovery (FEC), best for same-site or managed networks and prioritizes real-time performance over quality.
WAN -> BALANCED
*Designed for managed networks or high-quality internet connections with a balance of latency and stability. Moderate buffering, handles some jitter and packet loss, good default for most managed connections.
INTERNET -> ADAPTIVE
*Designed for unpredictable networks like marginal internet connections with high jitter. Larger jitter buffer, no packet recovery (FEC), best for public or variable networks.