diff --git a/docs/features/latency.md b/docs/features/latency.md index 3fda52f98..38ee40da3 100644 --- a/docs/features/latency.md +++ b/docs/features/latency.md @@ -41,37 +41,35 @@ that packet's declared scheduling time interval. SRT's approach to packet arrival time ===================================== -SRT provides two socket options `SRTO_PEERLATENCY` and `SRTO_RCVLATENCY`, -where the "latency" name was used for convenience (also for a common option -`SRTO_LATENCY`), but it doesn't mean that it will define the true time -distance between the `srt_sendmsg2` and `srt_recvmsg2` calls for the same -packet. This is only an extra delay added at the receiver side towards +SRT provides two socket options `SRTO_PEERLATENCY` and `SRTO_RCVLATENCY`. +While they have "latency" in their names, they do *not* define the true time +interval between the `srt_sendmsg2` and `srt_recvmsg2` calls for the same +packet. They are only used to add an extra delay (at the receiver side) to the time when the packet "should" arrive (ETS). This extra delay is used to -compensate two things: +compensate for two things: -* The extra network delay, that is, if the packet arrived later than it -"should have arrived" +* an extra network delay (that is, if the packet arrived later than it +"should have arrived"), or -* The packet retransmission, regarding that there might be a need to be -requested at least twice +* a packet retransmission. -Note that the values included in these formulas are values that are -actually present, but many of them are not controllable and many are -not even measurable. In many cases there are measured values that are -sums of other values, but ingredients can't be extracted. Values that -we get at the receiver side are actually two: +Note that many of the values included in these formulas are not controllable and +some cannot be measured directly. In many cases there are measured values +that are sums of other values, but the component values can't be extracted. -* ATS: actual arrival time. It's simply the time when the UDP packet +There are two values that we can obtain at the receiver side: + +* ATS: actual arrival time, which is the time when the UDP packet has been extracted through the `recvmsg` system call. * TS: time recorded in the packet header, set on the sender side and extracted from the packet at the receiver side Note that the timestamp in the packet's header is 32-bit, which gives -it more-less 2.5 minutes to roll over. Therefore there is the timestamp -rollover tracked and a segment increase is done in order to keep an +it more or less 2.5 minutes to roll over. Therefore timestamp +rollover is tracked and a segment increase is performed in order to keep an eye on the overall actual time. For the needs of the formula definitions -it will be stated that TS is the true difference between the connection +it must be stated that TS is the true difference between the connection start time and the time when the sending time has been declared when the sender application is calling any of the `srt_send*` functions (see [`srt_sendmsg2`](../API/API-functions.md#srt_sendmsg2) for details). @@ -82,28 +80,28 @@ SRT latency components To understand the latency components we need also other definitions: -* ETS: expected arrival time. This is the time of the packet when it -"should" arrive according to its TS +* **ETS** (Expected Time Stamp): The packet's expected arrival time, when it +"should" arrive according to its timestamp -* PTS: packet's play time. It's the time when SRT gives up the packet +* **PTS** (Presentation Time Stamp): The packet's play time, when SRT gives the packet to the `srt_recvmsg2` call (that is, it sets up the IN flag in epoll and resumes the blocked function call, if it was in blocking mode). -* STS: the time declared as the sending time when the packet was +* **STS** (Sender Time Stamp): The time when the packet was scheduled for sending at the sender side (if you don't use the -declared time, by default it's the monotonic time taken when this -function was called), which is represented by TS. +declared time, by default it's the monotonic time used when this +function is called). -* RTS: the same as STS, but calculated at the receiver side. The +* **RTS** (Receiver Time Stamp): The same as STS, but calculated at the receiver side. The only way to extract it is by using some initial statements. The "true latency" for a particular packet in SRT can be simply defined as: * `TD = PTS - STS` -Note that this is a stable definition (independent on the packet), +Note that this is a stable definition (independent of the packet), but this value is not really controllable. So let's define the PTS -for the packet `x`: +for a packet `x`: * `PTS[x] = ETS[x] + LATENCY + DRIFT` @@ -116,12 +114,12 @@ These components undergo the following formula: * `ETS[x] = start_time + TS[x]` -Note that it's not possible to simply define it basing on STS -because sender and receiver are two different machines that can only -see one another through the network, but their clocks are separate, -and can even run on different or changing speeds, while the only -visible phenomena happen only at a packet arrival machine. This -above formula, however, allows us to define the start time because +Note that it's not possible to simply define a "true" latency based on STS +because the sender and receiver are two different machines that can only +see one another through the network. Their clocks are separate, +and can even run at different or changing speeds, and the only +visible phenomena happen when packets arrive at the receiver machine. +However, the formula above does allow us to define the start time because we state the following for the very first data packet: * `ETS[0] = ATS[0]` @@ -140,10 +138,9 @@ is that there doesn't exist the common clock base for two network-communicating machines, so these components should be treated as something that does exist, but isn't exactly measurable). -But even if there is still this formula for ATS, it doesn't -apply to the real latency - this one is based strictly on ETS. -But you can apply this formula for the very first arriving -packet, because for this one they are equal: `ATS[0] = ETS[0]`. +This formula for ATS doesn't apply to the real latency, which is based strictly +on ETS. But you can apply this formula for the very first arriving packet, +because in this case they are equal: `ATS[0] = ETS[0]`. Therefore this formula is true for the very first packet: @@ -161,10 +158,10 @@ we have then: * `start_time = prg_delay[0] + net_delay[0] + snd_connect_time` -Note important thing: `start_time` is not the time of arrival of the first packet, +**IMPORTANT**: `start_time` is not the time of arrival of the first packet, but that time taken backwards by using the delay already recorded in TS. As TS should represent the delay towards `snd_connect_time`, `start_time` should be simply the same -as `snd_connect_time`, just on the receiver side, and so obviously shifted by the +as `snd_connect_time`, just on the receiver side, and so shifted by the first packet's delays of `prg_delay` and `net_delay`. So, as we have the start time defined, the above formulas: @@ -172,7 +169,7 @@ So, as we have the start time defined, the above formulas: * `ETS[x] = start_time + TS[x]` * `PTS[x] = ETS[x] + LATENCY + DRIFT` -define now the packet delivery time as: +now define the packet delivery time as: * `PTS[x] = start_time + TS[x] + LATENCY + DRIFT` @@ -180,11 +177,11 @@ and after replacing the start time we have: * `PTS[x] = prg_delay[0] + net_delay[0] + snd_connect_time + TS[x] + LATENCY + DRIFT` -and for the formula of TS we get STS, so we replace it: +and from the TS formula we get STS, so we replace it: * `PTS[x] = prg_delay[0] + net_delay[0] + STS[x] + LATENCY + DRIFT` -so the true network latency in SRT we can get by moving STS to the other side: +We can now get the true network latency in SRT by moving STS to the other side: * `PTS[x] - STS[x] = prg_delay[0] + net_delay[0] + LATENCY + DRIFT` @@ -198,35 +195,30 @@ value of the Arrival Time Deviation: * `ATD[x] = ATS[x] - ETS[x]` -The drift is then formed as: +The drift is then calculated as: * `DRIFT[x] = average(ATD[x-N] ... ATD[x])` -The value of the drift is tracked by appropriate number of samples and if +The value of the drift is tracked over an appropriate number of samples. If it exceeds a threshold value, the drift value is applied to modify the base time. However, as you can see from the formula for ATD, the drift is -simply taken from the real time when the packet was arrived, and the time -when it would arrive, if the `prg_delay` and `net_delay` values were +simply taken from the actual time when the packet arrived, and the time +when it would have arrived if the `prg_delay` and `net_delay` values were exactly the same as for the very first packet. ATD then represents the changes in these values. There can be two main factors that could result -in having this value as nonzero: - -1. There has been observed a phenomenon in several types of networks that -the very first packet arrives very quickly, but then as the data packets -come in regularly, the network delay slightly increases and then stays -for a long time with this increased value. This phenomenon could be -mitigated by having a reliable value of RTT, so once it's observed as -increased, a special factor could be used to decrease the positive value -of the drift, but this currently isn't implemented. This phenomenon also -isn't observed in every network, especially in a longer distance. - -2. The clock speed on both machines isn't exactly the same, which means -that if you decipher the ETS basing on the TS, after time it may result -in values that even precede the STS (and this way suggesting as if the -network delay was negative) or having an enormous delay (with ATS exceeding -PTS). This is actually the main reason of tracking the drift. - - - - - +in having this value as non-zero: + +1. A phenomenon has been observed in several types of networks where +the very first packet arrives quickly, but as subsequent data packets +come in regularly, the network delay slightly increases and then remains fixed +for a long time at this increased value. This phenomenon can be +mitigated by having a reliable value for RTT. Once the increase is observed +a special factor could be applied to decrease the positive value +of the drift. This isn't currently implemented. This phenomenon also +isn't observed in every network, especially those covering longer distances. + +2. The clock speed on both machines (sender and receiver) isn't exactly the same, +which means that if you decipher the ETS basing on the TS, over time it may result +in values that even precede the STS (suggesting a negative network delay) or that +have an enormous delay (with ATS exceeding PTS). This is actually the main reason +for tracking the drift.