JP2005510816A - Data transfer method - Google Patents

Abstract

The hybrid parallel / serial bus interface of the user equipment (UE) has a data block separator (40). The data block separator has an input configured to receive a data block and separates the data block into a plurality of nibbles. For each nibble, a parallel to serial converter (42 _i ) converts the nibble into serial data. One line transfers each nibble's serial data. The serial / parallel converter (46 _i ) converts the serial data of each nibble and reproduces the nibble. A data block reconstructor (48) combines the replayed nibbles into data blocks.

Description

  The present invention relates to bus data transfer. In particular, the present invention relates to reducing the number of lines for transferring bus data.

  An example of a bus used for transferring data is shown in FIG. FIG. 1 illustrates a receive and transmit gain controller (GC) 30, 32, and a GC controller 38 for use in a wireless communication system. A communication station such as a base station or user equipment transmits (TX) and receives (RX) a signal. In order to control the gain of these signals, the GCs 30, 32 adjust the gain of the received and transmitted signals so that they are within the operating range of the other receiving / transmitting components.

  A GC controller 38 is used to control the gain parameters for the GCs 30 and 32. As shown in FIG. 1, the GC controller 38 uses a power control bus, such as a 16-line bus 34, 36, to send gain values to the 8-line TX 36 and RX 34, respectively. The power control bus lines 34 and 36 enable high-speed data transfer, but in an integrated circuit (IC) such as an ASIC (Application Specific IC), many pins are provided to the GCs 30 and 32 and the GC controller 38. Or many connections between the GC 30, 32 and the GC controller 38. Increasing the number of pins requires additional circuit board area and connections. Increasing the number of IC connections consumes valuable IC space. A large number of pins or connections may increase the cost of the bus depending on the mounting method.

  Therefore, it would be desirable to have other solutions for data transfer.

  The hybrid parallel / serial bus interface has a data block separator. The data block separator has an input configured to receive a data block and separates the data block into a plurality of nibbles. For each nibble, a parallel to serial converter converts the nibble into serial data. One line transfers the serial data of each nibble. The serial / parallel converter converts the serial data of each nibble and reproduces the nibble. The data block reconstruction device combines the replayed nibble into data blocks.

  FIG. 2 is a block diagram of the hybrid parallel / serial bus interface, and FIG. 3 is a flowchart of data transfer of the hybrid parallel / serial bus interface. The data block is transferred from node 1 (50) to node 2 (52) through interface 44 (step 54). The data block separator 40 receives the block, separates the block into i nibbles, and transfers them on the i data transfer lines 44 (step 56). The value of i is based on a trade-off between the number of connections and the transfer rate. One solution for determining i is to first determine the maximum latency allowed to transfer the data block. Based on the maximum latency allowed, the minimum number of lines required to transfer the block is determined. With the minimum number of lines, the lines used to transfer data are selected to be at least that minimum number. Lines 44 are pins and may be their associated connection on the circuit board or connection on the IC. One solution for separating nibbles is to divide the block into nibbles from top to bottom. As shown in FIG. 4, an 8-bit block transfer on two lines, the block is separated into a 4-bit most significant nibble and a 4-bit least significant nibble.

  Another solution is to interleave the blocks across i nibbles. The first i bits of the block will be the first bit of each nibble. The second i bit becomes the second bit of each nibble and so on until the last i bit. The first bit is mapped to the first bit of the first nibble, as FIG. 5 shows an 8-bit block with two connections. The second bit is mapped to the first bit of the second nibble. The third bit is mapped to the second bit of the first nibble and so on until the last bit is mapped to the last bit of the second nibble.

  Each nibble is sent to a corresponding one of i parallel-to-serial (P / S) converters 42, converted from parallel bits to serial bits (step 58), and transferred serially through that line (step 58). Step 60). At the other end of each line is a serial / parallel (S / P) converter 46. Each S / P converter 46 converts the transmitted serial data into the original nibble (step 62). The i regenerated nibbles are processed by the data block reconstructor 48 to reconstruct the original data block (step 64).

  In another bidirectional solution, as shown in FIG. 6, data is transferred in both directions using i connections. Information data can be transferred in both directions, or information can be sent in one direction, and confirmation can be sent back in the other direction. The data block to be transferred from node 1 (50) to node 2 (52) is received by the data block separation and reconstructor 66. The separation / reconstruction device 66 separates the block into i nibbles. The i P / S converters 68 convert each nibble into serial data. A set of demultiplexers DM (MUXs / DMUXs) 71 connects each P / S converter 68 to a corresponding one of i lines 44. At node 2 (52), another set of DMs 75 connects line 44 to a set of S / P converters 72. The S / P converter 72 converts each received nibble serial data into the transmitted original nibble. The received nibble is reconstructed into the original data block by the data block separation / reconstruction device 76 and output as the received data block.

  For blocks transferred from node 2 (52) to node 1 (50), the data block is received by data block separation and reconstructor 76. The block is separated into nibbles, which are sent to a set of P / S converters 74. A P / S converter 74 converts each nibble to serial format for transfer through i lines 44. A set of DMs 75 in node 2 connects P / S converters 74 to i lines 44, and a set of DMs 71 in set of nodes 1 connects lines 44 to i S / P converters 70. To do. The S / P converter 70 converts the transmitted data into the original nibble. The data block separation reconstructor 66 reconstructs the data block from the received nibble and outputs the received data block. Since data is only sent in one direction at a time, this embodiment operates in half-duplex mode.

  FIG. 7 is a simplified diagram of one embodiment of a bidirectional switching circuit. The serial output from the P / S converter 68 of the node 1 is input to the [control terminal] of the tristate buffer 78 in the three state. Buffer 78 has another input connected to a voltage representing a high state. The output of buffer 78 is serial data and is sent over line 85 to tristate buffer 84 at node 2. A resistor 86 is connected between line 85 and ground. The buffer 84 at the node 2 passes the serial data to the S / P converter 72 at the node 2. Similarly, the serial output from the P / S converter 74 at the node 2 is input to the tristate buffer 82. Buffer 82 also has another input connected to the high state voltage. The serial output of buffer 82 is sent on line 85 to tristate buffer 80 at node 1. The buffer 80 of the node 1 passes the serial data to the S / P converter 70 of the node 1.

  In another embodiment, a portion of i lines 44 transfers data in one direction, and another line of i lines 44 transfers data to the other. At node 1 (50), the data block is received for transmission to node 2 (52). Based on the data throughput rate required for the block and the traffic demand in the opposite direction, j connections between 1 and i are used for the transfer of the block. The block is decomposed into j nibbles and converted into j sets of serial data using j of the i P / S converters 68. The corresponding number of j node 2 S / P converters 72 and the data block separation reconstructor 76 at node 2 regenerate the data block. In the opposite direction, i-j or k lines are used to transfer the block data.

  In a preferred embodiment of the bidirectional bus used in the gain control bus, the gain control value is sent in one direction and a confirmation signal is returned. The gain control value is sent alternately in one direction and the state of the gain control device is sent alternately in the other direction.

  One embodiment of a hybrid parallel / serial interface is performed in a synchronization system and is described in connection with FIG. A synchronization clock is used to synchronize the timing of the various components. A start bit is sent to indicate the start of a data block transfer. As shown in FIG. 8, each line is normally at a zero level. A start bit is sent to indicate the start of block transfer. In this example, all lines send a start bit, even though it is sufficient to send the start bit on one line. If the start bit is sent as a value on any line, the receiving node recognizes that the block data transfer has started. Each serial nibble is sent through the corresponding line. After a nibble transfer, all lines return to a normal state, such as a low state.

In another embodiment, the start bit is used as an indication of the function to be performed. One such embodiment is shown in FIG. As shown in FIG. 10, if the first bit of any connection is 1, the receiving node recognizes that block data is being transferred. As shown in the table of FIG. 11 as an example of the GC controller, three combinations of “01”, “10”, and “11” are used as combinations of start bits. “00” indicates that no start bit was sent. Each combination represents a function. In this example, “01” indicates that the subtraction function should be performed relatively, such as decreasing the data block value by one. “10” indicates that the addition function should be performed relatively, such as increasing the data block value by one. “11” indicates an absolute value function, and the block maintains the same value. Additional bits are used to increase the number of functions available. For example, up to 7 functions can be associated with two start bits for one line, or up to i ^{n + 1} −1 functions with ⁿ start bits for i lines. The processing unit 86 performs its function on the received data block as indicated by the start bit.

In another embodiment shown in FIG. 12, the start bit indicates the destination device. In FIG. 13, which shows an example of two destination devices / two lines, the start bit combination is associated with the destination device 88-92 of the data block being transferred. “01” is not used, “10” represents the device 2, and “11” represents the device 1. After the data block reconstructor 48 receives the start bit, the reconstructed block is sent to the corresponding device 88-92. Additional start bits can be used to increase the number of possible destination devices. With n start bits for each of the i lines, up to i ^{n + 1} −1 devices can be selected.

As shown in the table of FIG. 14, the start bit can be used to represent both a function and a destination device. FIG. 14 shows a three-connection system with two devices such as RXGC and TX · GC. Using the start bit for each line, three functions for the two devices are shown. In this example, the start bit on line 3 represents the destination device and is “0” for device 1 and “1” for device 2. The bits on lines 1 and 2 represent the function to be performed. “11” represents an absolute value function, “10” represents a relative increase function, and “01” represents a relative decrease. All three start bits are zero, ie, “000” is normally a non-data transfer state, and “001” is not used. Additional bits can be used to add more functions or devices. With n start bits for each of the i lines, up to i ^{n + 1} −1 function / device combinations are possible.

  FIG. 15 is a block diagram of a system that implements a start bit indicating both a function and a destination device. The regenerated nibble is received by the data block reconstructor 48. Based on the received start bit, the processor 86 performs the indicated function and the processed block is sent to the indicated destination device 88-92.

  As shown in the flowchart of FIG. 16, a start bit indicating function / destination is added to each nibble (step 94). The nibble is sent over i lines (step 96). Using the start bit, the appropriate function is performed on the data block, the data block is sent to the appropriate destination, or both are performed (step 98).

  In order to increase its throughput in a synchronization system, both rising (even) and falling (odd) edges of the clock are used to transfer block data. One embodiment is shown in FIG. Data blocks are received by data block separator 100 and separated into two sets of i nibbles (even and odd). Each set of i nibbles is sent to each of a set of i P / S devices 102, 104. As shown in FIG. 17, the odd-numbered P / S devices 102 have i P / S devices and have a clock signal inverted by an inverter 118. As a result, the inverted clock signal is delayed by half a cycle with respect to the system clock. A set of i MUXs 106 selects between an even number of P / S devices 104 and an odd number of P / S devices 102 at twice the clock rate. As a result, the data transferred over each connection is twice the clock rate. The other side of each connection is a corresponding DEMUX 108. The DEMUX 108 connects each line 44 to the even buffer 112 and the odd buffer 110 sequentially at a double clock rate. Each buffer 112, 110 receives the corresponding even and odd bits and holds their values for one clock cycle. The even and odd sets of S / P devices 116, 114 play even and odd nibbles. Data block reconstructor 122 reconstructs data blocks from the transferred nibbles.

  FIG. 18 shows the data transfer through the lines of the system using the rising and falling edges of the clock. Even and odd data transferred by line 1 are shown. The shaded portion indicates falling clock data in the combined signal, and the non-shaded portion indicates rising clock data. As shown, the data transfer rate is doubled.

  FIG. 19 is a preferred embodiment of a hybrid parallel / serial interface used between the GC controller 38 and the GC 124. A data block having 16-bit GC control data (8-bit RX and 8-bit TX) is sent from the GC controller 38 to the data block separator 40. The data block is separated into two nibbles, such as two 8-bit nibbles. A start bit is added to each nibble so that there are 9 bits per nibble. Two nibbles are transferred over two lines using two P / S converters 42. When detecting the start bit, the S / P converter 46 converts the received nibble into a parallel format. The data block reconstructor reconstructs the original 16 bits and controls the gain of the GC 124. As shown in FIG. 11, if the function is indicated by a start bit, the AGC 124 performs the function on the received block prior to gain adjustment.

  FIG. 20 is another preferred embodiment of a hybrid parallel / serial interface that uses three lines between the GC controller 38 and the RXGC 30 and TXGC 32. The GC controller 38 sends the data block to the GCs 30, 32 along with the appropriate RX and TX gain values and the start bit of FIG. When the start bit of FIG. 14 is used, device 1 is RXGC 30 and device 2 is TXGC 32. The data block separator 40 separates a data block to be transferred by three lines into three nibbles. Using three P / S converters 42 and three S / P converters 46, the nibble is transferred serially over the line and converted to the original nibble. Data block reconstructor 48 reconstructs the original data block and performs functions as indicated by the start bit, such as relative increase, relative decrease and absolute value. The resulting data is sent to either RX or TXGC 30, 32 as indicated by the start bit.

It is a figure which shows RXGC, TXGC, and GC controller. FIG. 3 is a block diagram of a hybrid parallel / serial bus interface. FIG. 6 is a flowchart of data block transfer using a hybrid parallel / serial bus interface. FIG. 5 shows the separation of blocks into the highest and lowest nibbles. FIG. 4 is a diagram illustrating separating blocks using data interleaving. FIG. 2 is a block diagram of a bi-directional hybrid parallel / serial bus interface. It is a figure of one Example of one bidirectional line. FIG. 5 is a timing diagram showing start bits. It is a block diagram of a hybrid parallel / serial bus interface capable of function control. FIG. 7 is a timing diagram of start bits for a hybrid parallel / serial bus interface capable of function control. 6 is a table of one embodiment of start bits indicating functions. 2 is a block diagram of a hybrid parallel / serial bus interface that controls a destination. FIG. It is the table of one Example of the start bit which shows a destination. FIG. 6 is a table of one embodiment of start bits indicating destination / function. FIG. FIG. 3 is a block diagram of a hybrid parallel / serial bus interface that controls destination / function. FIG. 6 is a flowchart of a start bit indicating a destination / function. FIG. FIG. 3 is a block diagram of a hybrid parallel / serial bus interface that uses rising and falling edges of a clock. FIG. 5 is a timing diagram of a hybrid parallel / serial bus interface using rising and falling clock edges. 2 is a block diagram of a two line GC / GC controller bus. FIG. 3 is a block diagram of a three line GC / GC controller bus. FIG.

Claims (20)

A method used by a user equipment (UE) to transfer data comprising:
Supplying data blocks,
Separating the data block into a plurality of nibbles each having a plurality of bits;
For each nibble,
Converting the nibble into serial data;
Providing a line and transferring the nibble serial data over the line;
Converting the nibble serial data into parallel data and reproducing the nibble; and
The reproduced nibble is combined with the data block.   The method of claim 1, wherein the number of bits in the data block is N, the number of lines is i, and 1 <i <N.   The method of claim 1, wherein the number of bits in a nibble is four and the number of lines is two. A method used by a user equipment (UE) for transferring a data block through an interface connecting a first node to a second node comprising:
Separating the data block into m sets of n bits;
Adding a start bit to each of the m sets, wherein the m start bits collectively represent one particular mathematical function or destination;
Transferring each of the m sets from the first node by a separate line;
Receiving each of the forwarded m sets at the second node; and
Using the received m sets according to the m start bits.   At least one of the m start bits is in a 1 state and all the individual lines are in a 0 state when the interface is not transmitting data. Item 5. The method according to Item 4.   The method of claim 4, wherein the m start bits indicate the start of data transfer.   The method of claim 4, wherein the m start bits collectively represent a particular mathematical function rather than a destination.   The method of claim 4, wherein the functions collectively represented by the m start bits include a relative increase, a relative decrease, and an absolute value function.   The method of claim 4, wherein the m start bits collectively represent a specific destination rather than a mathematical function.   The method of claim 9, wherein destinations collectively represented by the m start bits include RX and TX gain controllers.   The method of claim 4, wherein the m start bits collectively represent both a specific mathematical function and a specific destination. A method used by a user equipment (UE) to determine the number of i bus connections needed to transfer block data over the bus, each block of the block data comprising a number of bits N,
Determining a maximum latency allowed for the transfer of the block data;
Determining the minimum number of connections required to transfer the data block with the maximum latency; and
determining i, which is at least the minimum number of connections required.   The method of claim 12, wherein the i bus connections correspond to i pins on a chip.   14. The method of claim 13, wherein 1 <i <N. Generating a data block by a controller of a gain controller (GC), said data block having n bits representing a gain value;
Transferring the data block from the GC controller to the GC via i (1 <i <n) lines;
Receiving the data block at the GC; and
Adjusting the gain of the GC using the gain value of the data block. A method used by a user equipment (UE), comprising: Prior to transferring the data block, separating the data block into a plurality of nibbles and transferring each nibble by a different line of the lines; and
16. The method of claim 15, further comprising combining the nibble with the data block after receiving the data block.   The method of claim 16, wherein a start bit is added to each nibble.   The method of claim 17, wherein the start bit indicates a function to be performed.   16. The method of claim 15, wherein the mathematical function indicated by the start bit includes a relative increase, a relative decrease, and an absolute value function.   16. The method of claim 15, wherein the GC includes RXGC and TXGC, and the start bit indicates whether the data block is sent to the RXGC or TXGC.

Images