Compact MPE-FEC erasure location cache memory for DVB-H receiver
View Patent ↗Location cache memory architectures that only require 32 Kbits or less per frame to store erasure information with simple address mapping to the main MPE-FEC RAM for easy column-wise and row-wise access. Alternative architectures are designed to greatly reduce the size and logic complexity of the MPE-FEC erasure cache memory. Two architectures reduce the erasure cache size down to 32 Kbits and 28 Kbits, correspondingly, without introducing additional erasure locations, while another architecture further reduces the required memory size down to 16K, 8K, 4K, or 2K bits with a slight increase in the total erasure locations. All architectures group the data in MPE-FEC frame memory into blocks of 2 M consecutive bytes and use one or a few bits to store the erasure status in each block, thereby, greatly reducing the required cache memory size.
1. A method of identifying erasure locations in transmitted data received by a Digital Video Broadcasting over Handheld (DVB-H) receiver, said method comprising:
grouping data bytes from Internet Protocol (IP) datagrams in a Multi-Protocol Encapsulator Forward Error Correction Random Access Memory (MPE-FEC RAM) unit in a column-wise manner into data blocks comprising 2 M consecutive bytes per data block, wherein M comprises a positive integer;
using an erasure cache RAM comprising a size of 2 (10−M) ×255 entries having (M+2) bits per entry to indicate an erasure status of each data byte in a MPE data frame;
using M bits in the (M+2)-bit entry in said erasure cache RAM to identify a 2 M −1 potential boundary of said IP datagrams when a corresponding data block covers two said IP datagrams; and
using a remaining two bits in said (M+2)-bit entry to identify potential erasure location conditions in said IP datagrams.
2. The method of claim 1 , wherein said MPE-FEC RAM comprises 2 (10−M) ×255 data blocks per said MPE data frame.
3. The method of claim 1 , wherein 2 M is no larger than a smallest IP datagram length such that each said data block of 2 M bytes comprises data from at most two said IP datagrams.
4. The method of claim 1 , wherein a size of said erasure cache RAM comprises at most 32 Kbits.
5. The method of claim 1 , wherein said potential erasure location conditions comprises:
a first condition where both of said two said IP datagrams are error free;
a second condition where only a first one of said two said IP datagrams has errors;
a third condition where only a second one of said two said IP datagrams has errors; and
a fourth condition where said both of said two IP datagrams have errors.
6. The method of claim 5 , wherein in said first condition, no erasures exist in the 2 M byte data block.
7. The method of claim 5 , wherein in said second condition, all said data bytes before a corresponding IP datagram boundary are marked as erasures.
8. The method of claim 5 , wherein in said third condition, all said data bytes after a corresponding IP datagram boundary are marked as erasures.
9. The method of claim 5 , wherein in said fourth condition, all said data bytes in a corresponding data block is marked as an erasure.
10. The method of claim 1 , wherein said M bits in said (M+2)-bit entry is used to indicate whether a corresponding data block covers data from exactly one IP datagram.
11. The method of claim 5 , wherein said two bits in said (M+2)-bit entry is used to indicate the first, second, third, and fourth conditions.
12. A method of identifying erasure locations in transmitted data received by a Digital Video Broadcasting over Handheld (DVB-H) receiver, said method comprising:
grouping data bytes from Internet Protocol (IP) datagrams in a Multi-Protocol Encapsulator Forward Error Correction Random Access Memory (MPE-FEC RAM) unit in a column-wise manner into data blocks comprising 2 M consecutive bytes per data block, wherein M comprises a positive integer;
using an erasure cache RAM comprising a size of 2 (10−M) ×255 entries having (M+1) bits per entry to indicate an erasure status of each data byte in a MPE data frame;
using M bits in the (M+1)-bit entry in said erasure cache RAM to identify a 2 M −1 potential boundary of said IP datagrams when a corresponding data block covers two said IP datagrams with a different cycle redundancy check (CRC) matching status; and
using a remaining one bit in said (M+1)-bit entry to identify potential erasure location conditions in said IP datagrams.
13. The method of claim 12 , wherein said MPE-FEC RAM comprises 2 (10−M) ×255 data blocks per said MPE data frame.
14. The method of claim 12 , wherein 2 M is no larger than a smallest IP datagram length such that each said data block of 2 M bytes comprises data from at most two said IP datagrams.
15. The method of claim 12 , wherein a size of said erasure cache RAM comprises at most 28 Kbits.
16. The method of claim 12 , wherein said potential erasure location conditions comprises:
a first condition where only a first one of said two said IP datagrams has errors; and
a second condition where only a second one of said two said IP datagrams has errors.
17. The method of claim 16 , wherein in said first condition, all said data bytes before a corresponding IP datagram boundary are marked as erasures.
18. The method of claim 16 , wherein in said second condition, all said data bytes after a corresponding IP datagram boundary are marked as erasures.
19. The method of claim 12 , wherein said M bits in said (M+1)-bit entry is used to indicate whether a corresponding data block covers data from exactly one IP datagram or covers two datagrams of the same CRC matching status.
20. The method of claim 16 , wherein said remaining one bit in said (M+1)-bit entry is used to indicate the first and second conditions for a corresponding data block.
21. A method of identifying erasure locations in transmitted data received by a Digital Video Broadcasting over Handheld (DVB-H) receiver, said method comprising:
grouping data bytes from Internet Protocol (IP) datagrams in a Multi-Protocol Encapsulator Forward Error Correction Random Access Memory (MPE-FEC RAM) unit in a column-wise manner into data blocks comprising 2 M consecutive bytes per data block, wherein M comprises a positive integer;
using an erasure cache RAM comprising a size of 2 (10−M) ×255 entries having only one bit per entry to indicate an erasure status of each data byte in an entire data block; and
using said only one bit to identify potential erasure location conditions in said IP datagrams.
22. The method of claim 21 , wherein when a particular data block covers more than one IP datagram and there are cycle redundancy check (CRC) mismatches in any of the IP datagrams, then said particular data block is marked as an erasure.
23. The method of claim 22 , wherein when said particular data block is not marked as an erasure, then said particular data block is marked as error free.
24. The method of claim 21 , wherein when said M equals 4, a size of said erasure cache RAM per MPE frame equals 16K.
25. The method of claim 21 , wherein when said M equals 5, a size of said erasure cache RAM per MPE frame equals 8K.
26. The method of claim 21 , wherein when said M equals 6, a size of said erasure cache RAM per MPE frame equals 4K.
27. The method of claim 23 , further comprising using said only one bit to identify two possible erasure statuses of a corresponding data block.
28. A lossless method of identifying erasure locations in transmitted data received by a Digital Video Broadcasting over Handheld (DVB-H) receiver, said method comprising:
grouping data bytes from Internet Protocol (IP) datagrams in a Multi-Protocol Encapsulator Forward Error Correction Random Access Memory (MPE-FEC RAM) unit in a column-wise manner into data blocks comprising 2 M consecutive bytes per data block, wherein M comprises a positive integer;
using an erasure cache RAM comprising a size of 2 (10−M) ×255 entries having (M+1) bits per entry to indicate an erasure status of each data byte in a MPE data frame; and
using (M+1)-bit binary encoding to represent 2 (M+1) erasure patterns, wherein the erasure patterns occur when a corresponding data block covers one or two of said IP datagrams.