Method and apparatus for increased quality of voice transmission over the internet

• US 5,883,891 A
• Filed: 04/30/1996
• Issued: 03/16/1999
• Est. Priority Date: 04/30/1996
• Status: Expired due to Fees

• Current Assignee: iCall Incorporated, Mailsignal, Inc.
• Original Assignee: Joel Libove, Wyatt Williams
• Inventors: Williams, Wyatt, Libove, Joel
• Primary Examiner(s): Nguyen, Chau
• Assistant Examiner(s): Hoosain, Allan
• Application Number:
  US08/641,239
• Time in Patent Office: 1,050 Days
• Field of Search: 370/352, 370/353, 370/356, 370/380, 370/389, 370/394, 370/396, 370/401, 370/473, 370/485, 370/427, 370/435, 370/450, 370/465, 370/37.01, 370/37.02, 370/37.12, 379/90.01, 379/93.01, 379/93.05, 379/93.09, 379/100.15, 379/100.16
• US Class Current: 370/356
• CPC Class Codes: H04L 12/64 Hybrid switching systems
• CPC Class Codes: H04L 45/24 Multipath

Abstract

To satisfactorily transmit two-way speech on digital computer networks, such as the Internet, it is necessary to reduce unpredictable delays due to variations in loading, routing and other factors. The present invention transmits extra packets of specially computed data along with the data representing speech. These extra packets are used at the receiving server to reconstruct any missing or late data packets. In one embodiment, any single packet of missing data in a stream of packets may be reconstructed. In additional embodiments, multiple packets of missing data may be reconstructed. This method is particularly suited to eliminating or reducing delays in networks in which data is sent down multiple diverse paths. Unlike prior art, it is not necessary to send full sets of redundant data down these diverse paths. The present invention significantly reduces latency with only a fraction of the added network bandwidth required with prior art methods that use full sets of redundant data.

Claims

1. A method of transmitting a caller'"'"'s voice over the Internet from a source node to a caller'"'"'s destination node, including the steps of converting the caller'"'"'s voice into digital data, formatting the data into bit data packets, transmitting the data packets over the Internet and converting the data packets into sound which is a replica of the caller'"'"'s voice;

characterized in that the method includes the steps of;

(a) generating at the source node and transmitting a plurality of replication packets each based on the bits of a group of the data packets, the replication packets being transmitted over the Internet to the destination node along with the group of data packets;

(b) selecting a plurality of routes, each route having a sequence of intermediate nodes, and formatting a header of each data packet with a sequence of intermediate node addresses to enable each packet to be transmitted over one of the selected routes, and transmitting data packets of the caller'"'"'s voice over different routes of the plurality of routes;

(c) at the destination node, detecting if any data packets are missing from the group of data packets and exactly replicating any missing data packets by computer computations using the replication packet and non-missing data packets of the group of data packets to generate replicated data packets which exactly replicate the missing packets; and

(d) using the replicated data packets along with the other data packets of the group of data packets to reproduce the caller'"'"'s voice.

2. The method of claim 1 wherein the conversion of the caller'"'"'s voice is by a source node server computer connected by a communication line to the Internet and the method includes, at the source node formatting a group of data packets as a 2-dimensional or 3-dimensional matrix in the form of rows and columns and generating the replication packet as a row of the matrix.

3. The method of claim 2 and generating the replication packet by determining if the sum of the aligned bits of each column is an even or odd number.

4. The method of claim 1 and, at the destination node, forming each data packet group into a 2-dimensonal or 3-dimensional matrix of data packets and a plurality of replication packets so that more than one missing packet may be replicated.

5. The method as in claim 1 and transmitting an echo packet from the source node to the destination node and recording in the echo packet the address and time it is communicated at each Internet intermediate node.

6. The method of claim 5 and transmitting the recorded echo packet from the destination node to the source node.

7. The method of claim 5 and transmitting a plurality of echo packets over a plurality of routes of the Internet and determining the best available routes based on a comparison between the plurality of echo packets after said transmission.

8. The method of claim 1 wherein the caller calls from a telephone, the telephone is connected to the source node, the caller speaks to a person being called via a telephone connected to the destination mode, and the method includes the steps of claim 1 executed as to the voice of the person called so that the destination node becomes a source node and the source node becomes a destination node when the person called is speaking.

9. A system for transmitting a caller'"'"'s voice over the Internet including converter means for converting a caller'"'"'s voice into bits of digital data, and transmitting means for transmitting the data packets over the Internet, characterized in that the system includes:

(a) source node means for formatting and transmitting a replication packet derived from the bits of a group of data packets, the replication packet being transmitted over the Internet along with a group of data packets;

(b) means for selecting a plurality of Internet routes, each route comprising a source node;

a destination node and a sequence of intermediate nodes;

(c) header formatting means for formatting a header for each packet with a sequence of intermediate node addresses to enable each packet to be transmitted over one of the selected routes;

(d) transmitting means for transmitting different packets representing the caller'"'"'s voice over different routes of the selected routes;

(e) destination node means for detecting if any data packets are missing from the group of data packets and replicating the missing data packets by computations using the replication packet and non-missing data packets of the group of data packets to generate replicated data packets; and

(f) sound means for reproducing the caller'"'"'s voice using the replicated data packets along with the non-missing data packets of the group.

10. The system of claim 9 and means in (b) for forming a group of data packets as a 2-dimensional matrix in the form of rows and columns and forming the replication packet as a row of the matrix.

11. The system of claim 10 and means in (b) for forming the replication packet by determining if the sum of the aligned bits for each column is an even or odd number.

12. The system of claim 9 and means in (b) for forming each data packet group into a 2-dimensional or 3-dimensional matrix of data packets and a plurality of replication packets so that more than one missing packet may be replicated.

13. The system as in claim 9 and means for transmitting an echo packet form the source node to the destination mode;

and recording means for recording on the echo packet the address and the time packet is transmitted from each Internet node.

14. The system of claim 13 and means for transmitting the recorded echo packet from the destination node to the source node.

15. The system of claim 14 and means for transmitting the echo packets over a plurality of routes and determining the best available routes based on the plurality of transmitted echo packets, after they have traveled the routes.