METHOD AND APPARATUS FOR IMPROVING HTTP ADAPTIVE STREAMING PERFORMANCE USING TCP MODIFICATIONS AT CONTENT SOURCE

• US 20130308454A1
• Filed: 05/18/2012
• Published: 11/21/2013
• Est. Priority Date: 05/18/2012
• Status: Active Grant

• Current Assignee: RPX Corporation
• Original Assignee: Alcatel-Lucent Canada Incorporated (Nokia Corporation), Alcatel-Lucent USA, Inc. (Nokia Corporation)
• Inventors: Akhtar, Shahid, Chaudhry, Ather J., Craciun, Viorel, Liefooghe, Pieter J.
• Granted Patent: US 9,130,843 B2
• Time in Patent Office: Days
• US Class Current: 370/235
• CPC Class Codes: H04L 43/0864 Round trip delays
• CPC Class Codes: H04L 65/65 Network streaming protocols, e.g. real-time transport protocol [RTP] or real-time control protocol [RTCP]
• CPC Class Codes: H04L 65/80 Responding to QoS
• CPC Class Codes: H04L 67/02 based on web technology, e.g. hypertext transfer protocol [HTTP]
• CPC Class Codes: H04L 69/16 Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
• CPC Class Codes: H04L 69/163 In-band adaptation of TCP data exchange; In-band control procedures

Abstract

A method of sending hypertext transfer protocol (HTTP) adaptive streaming (HAS) content from a content source to a client over a transmission control protocol (TCP) connection between the content source and the client may include receiving a request for one or more first HAS data chunks from the client; sending the one or more first HAS data chunks to the client; generating idle delay values indicating an amount of delay in the TCP connection during an idle time period, the idle time period being a time period over which no HAS data chunks are being sent from the content source to the client; receiving a request for a second HAS data chunk from the client; setting the size of a starting TCP congestion window based on the idle delay values; and sending the second HAS chunk to the client using the starting TCP congestion window.

Claims

1. A method of sending hypertext transfer protocol (HTTP) adaptive streaming (HAS) content from a content source to a client over a transmission control protocol (TCP) connection between the content source and the client, the method comprising:

measuring network delay values of the TCP connection; and

determining a size of a starting TCP congestion window (cwnd) based on the measured delay values.

2. The method of claim 1, further comprising:

receiving a request for one or more first HAS data chunks from the client;

sending the one or more first HAS data chunks to the client;

receiving a request for a second HAS data chunk from the client; and

sending the second HAS chunk to the client using the determined cwnd,

wherein the network delay values include idle delay values indicating an amount of delay in the TCP connection during an idle time period, the idle time period is a time period over which no HAS data chunks are being sent from the content source to the client, and the starting cwnd is determined based on the idle delay values.

3. The method of claim 2 wherein generating the idle delay values includes:

generating one or more loopback messages and sending the one or more loop back messages to the HAS client during the idle time period;

measuring a plurality of first round trip times (RTTs) corresponding to each of the one or more loop back messages, respectively; and

generating the idle delay values based on the measured plurality of RTTs.

4. The method of claim 3 wherein the idle delay values include a short term idle delay value, and generating the short term idle delay value includes averaging RTTs falling in a first moving window of the plurality of first RTTs.

5. The method of claim 4 wherein the idle delay values further include a long term delay value, and generating the long term idle delay value includes averaging RTTs falling in a second moving window of the plurality of first RTTs, the first moving window including fewer RTT values than the second moving window.

6. The method of claim 3, further comprising:

generating an active delay value indicating an amount of delay in the TCP connection during an active time period, the active time period being a time period during which one or more HAS data chunks are being sent,

wherein setting the size of the starting TCP congestion window includes setting the size of a starting TCP congestion window based on the idle delay values and the active delay value.

7. The method of claim 6, wherein sending the one or more first HAS data chunks includes:

dividing the first one or more data chunks into a plurality of data segments, and

sending each of the plurality of data segments to the client; and

wherein generating the active delay value includes

measuring a plurality of second round trip times (RTTs) corresponding to each of the plurality of data segments, respectively, and

generating the active delay values based on the measured plurality of second RTTs.

11. The method of claim 3, wherein the one or more loopback messages are empty MP4 HAS messages.

8. The method of 7, wherein generating the active delay values based on the measured second RTTs includes averaging RTTs falling in a moving window of the plurality of second RTTs.

9. The method of 7, further comprising:

storing the n longest successful RTTs from among the first plurality of RTTs and the second plurality of RTTs, successful RTTs being RTTs of data segments or loopback messages for which the content source received no indication of loss; and

determining a lossless delay value based on an average of the n longest successful RTTs,

wherein setting the size of the starting TCP congestion window includes setting the size of a starting TCP congestion window based on the idle delay values, the active delay value, and the lossless delay value.

10. The method of 9, further comprising:

determining a fraction value based on the idle delay values, the active delay value, and the lossless delay value; and

determining whether the fraction value is greater than a reference value,

wherein setting the size of the starting TCP congestion window includes determining a product of the fraction value and a size of a last TCP congestion window used to send the one or more first HAS chunks to the client, and setting the determined product as the size of the starting TCP congestion window.

12. A network apparatus configured to send hypertext transfer protocol (HTTP) adaptive streaming (HAS) content to a client over a transmission control protocol (TCP) connection between the network apparatus and the client, the apparatus comprising:

a receiver unit configured to receive data;

a transmitting unit configured to transmit data;

a memory unit configured to store parameters corresponding with characteristics of the TCP connection; and

a processing unit coupled to the transmitting unit, the receiving unit, and the memory unit and configured to control operations associated with sending HAS content to the client, the operations including,

measuring network delay values of the TCP connection; and

determining a size of a starting TCP congestion window (cwnd) based on the measured delay values.

13. The network apparatus of claim 12, wherein the processor is further configured to control operations including,

receiving a request for one or more first HAS data chunks from the client;

sending the one or more first HAS data chunks to the client;

receiving a request for a second HAS data chunk from the client; and

sending the second HAS chunk to the client using the cwnd,

wherein the network delay values include idle delay values indicating an amount of delay in the TCP connection during an idle time period, the idle time period is a time period over which no HAS data chunks are being sent from the content source to the client, and the starting cwnd is determined based on the idle delay values.

14. The network apparatus of claim 13 wherein the processing unit is configured such that generating the idle delay values includes

generating one or more loopback messages and sending the one or more loop back messages to the HAS client during the idle time period;

measuring a plurality of first round trip times (RTTs) corresponding to each of the one or more loop back messages, respectively; and

generating the idle delay values based on the measured plurality of first RTTs.

15. The network apparatus of claim 14 wherein the processor is configured such that the idle delay values include a short idle delay value, and generating the short idle delay value includes averaging RTTs falling in a first moving window of the plurality of first RTTs.

16. The network apparatus of claim 15 wherein the processor is configured such that the idle delay values further include a long idle delay value, and generating the long idle delay value includes averaging RTTs falling in a second moving window of the plurality of first RTTs, the first moving window including fewer RTT values than the second moving window.

17. The network apparatus of claim 16, wherein the processor is further configured to control operations including

generating an active delay value indicating an amount of delay in the TCP connection during an active time period, the active time period being a time period during which one or more HAS data chunks are being sent,

wherein setting the size of the starting TCP congestion window includes setting the size of a starting TCP congestion window based on the idle delay values and the active delay value.

18. The network apparatus of claim 17, wherein the processor is configured such that sending the one or more first HAS data chunks includes

dividing the first one or more data chunks into a plurality of data segments, and

sending each of the plurality of data segments to the client;

and wherein generating the active delay value includes

measuring a plurality of second round trip times (RTTs) corresponding to each of the plurality of data segments, respectively, and

generating the active delay values based on the measured plurality of second RTTs.

22. The network apparatus of claim 14, wherein the one or more loopback messages are empty MP4 HAS messages.

19. The network apparatus of 18, wherein the processor is configured such that generating the active delay values based on the measured second RTTs includes averaging RTTs falling in a moving window of the plurality of second RTTs.

20. The network apparatus of 19, wherein the processor is further configured to control operations including

storing the n longest successful RTTs from among the first plurality of RTTs and the second plurality of RTTs, successful RTTs being RTTs of data segments or loopback messages for which the content source received no indication of loss;

and

determining a lossless delay value based on an average of the n longest successful RTTs,

wherein setting the size of the starting TCP congestion window includes setting the size of a starting TCP congestion window based on the idle delay values, the active delay value, and the lossless delay value.

21. The network apparatus of 20, wherein the processor is further configured to control operations including

determining a fraction value based on the idle delay values, the active delay value, and the lossless delay value;

and

determining whether the fraction value is greater than a reference value,

wherein setting the size of the starting TCP congestion window includes

determining a product of the fraction value and a size of a last TCP congestion window used to send the one or more first HAS chunks to the client and setting the determined product as the size of the starting TCP congestion window, if the fraction value is greater that the reference value, and

setting the size of the staring TCP congestion window in accordance with one the TCP standards defined by RFC 2861 and RFC 2581, if the fraction value is not greater that the reference value.