Related applications are:
“Flexible CDMA System Architecture”, Ser. No. ______;
“Generic Data Path Processor for W-CDMA Applications”, Ser. No. ______;
“Generic Finger Architecture for W-CDMA Applications”, Ser. No. ______.
Each of these applications is filed concurrently herewith and is incorporated herein by reference.
This invention relates generally to wireless communication systems.
Wireless communication has many applications in consumer and business markets. Among the many applications are: mobile wireless, fixed wireless, unlicensed Federal Communications Commission (FCC) wireless, local area network (LAN), cordless telephony, personal base station, telemetry, and others. Generally, each of these applications utilizes unique and frequently incompatible modulation techniques and protocols.
Wireless communication devices, such as cellular handsets, typically need to acquire certain cell specific information and characteristics before negotiating a service with a base station. For this purpose, each base station transmits certain cell specific information necessary for a user to acquire services such as paging or cellular telephony from the base station. For example, in CDMA systems, the cell specific information is contained in pilot and/or synchronization channels. The pilot and/or synchronization channels are spread and scrambled with cell specific pseudo-random noise or gold code sequences, which form the basis for frame, slot, and bit timing synchronization for a handset.
Multiple users are typically in communication with a single base station. Although such users operate on the same frequency at the same time, various techniques allow multiple users to be distinguished from one another. In CDMA systems, each handset user is assigned a different orthogonal code that is used to spread the data transmitted from the base station so as to distinguish it from the data transmitted to other handset users.
The front-end processor 104 is coupled to the antenna 102 to receive data. The front-end processor 104 includes components and performs functions that are known to those skilled in the art. These components are not shown in the front-end processor 104 for purposes of clarity.
After data has been processed by the front-end processor, the processed data is supplied to the-base band processor 106. In spread spectrum systems, the base band processor has to be able to identify, despread, and decode the data. Despreading (i.e., multiplication of the process data by the same binary spreading waveform as was used to spread the data at the transmitter) and removal of the carrier modulation restore the original baseband data waveform.
In practice, multiple copies of the same signal are typically received at communications device 100 within a short time of each other. These copies, which are sometimes called multipath components arise because the signals take different paths of different length from the transmitter antenna to the receiver antenna. In the case of a CDMA system, it is feasible and advantageous to despread and decode several of the multipath components, realign them so that they are also in phase and combine them to produce a stronger signal. To do this, the base band processor in a CDMA system typically takes the form of a rake receiver that has several fingers, each one of which is a receiver that despreads and decodes one of the multipath components. General information about rake receivers can be found at pages 972-982 of J. S. Lee, L. E. Miller, CDMA Systems Engineering Handbook (Artech House 1998).
Generally, received data is sampled at a rate known as the chipping rate. In the IS-95 and 3 GPP CDMA standards, the chipping rates are 1.2288 MHz and 3.84 MHz, respectively, which correspond to sampling periods of 0.814 microseconds and 0.2604 microseconds. The sampling period is known as a chip.
Data is usually processed in pairs, commonly referred to as “In-Phase” (I) and “Quadrature” (Q) data. A rake finger is supplied with three samples, typically labeled Early, On-Time, and Late, where each sample includes a pair of data (i.e., I and Q data). The On-Time sample contains the data to be decoded. The Early and Late samples are used in tracking tools to ensure that the On-Time sample represents the center of the chip.
Two examples of chipping rate are 1.2288 MHz and 3.84 MHz which correspond to sampling periods of 0.814 microseconds and 0.2604 microseconds. These are used in IS95 and 3GPP CDMA systems, respectively.
Data arriving at the base band processor 106 is typically over-sampled at an over-sampling rate (e.g., 4× or 8× over-sampling rate) measured with respect to the chipping rate. For example: if the chipping rate is 1.2288 MHz and the over-sampling rate is 4× then the samples will be at 4.9152 MHz. Directly passing received over-sampled data to the base band processor 106 is generally inefficient because the sampling rate of the data may not be at the optimal rate for processing by the base band processor 106 rate. Thus, some type of front-end storage is required. Because the over-sampling rate is typically very fast, such front-end storage is likely to be very expensive, and more memory storage capacity is required as the sampling rate increases. Thus, a need arises for apparatus and methods that provide fast storage of raw data at the front end of the baseband processor without requiring excessive memory space.
Further, data is typically despread by multiple rake fingers. A need arises for apparatus and methods that allow reuse of multiple rake fingers to conserve physical silicon space and allow the multiple rake fingers to access the data storage substantially concurrently to achieve faster processing.
In an exemplary embodiment of the present invention, the sampling rate of received data is decimated by a factor of D and the received data is stored in memory at the decimated rate. By storing received data at the decimated rate, storage requirements are reduced by a factor of D. This can be done because the data, prior to sampling, has been filtered such that all information at sampling rates greater than twice the filter cut-off frequency has been removed. The stored data is subsequently interpolated by a factor of IR to restore it to a faster rate, a multiple of the chipping rate. In an exemplary embodiment, decimation and interpolation operations occur at the beginning of the signal processing chain, thus providing the benefit of reduced storage requirements at the front end of the processing chain. However, decimation and interpolation operations can occur anywhere in the signal processing chain and therefore are not limited to implementations at the front end.
In an exemplary embodiment, the memory is a circular buffer comprising a set of registers. Each register is divided into two parts for storing in-phase and quadrature (I and Q) data, respectively. Once data has been stored into memory (at a decimated rate), a despreading circuit (e.g., a rake receiver), which includes multiple rake fingers, can access the data. The multiple rake fingers can substantially simultaneously access data in the memory via a selector circuit for each rake finger. Each selector circuit includes multiple multiplexers controlled by a set of select lines. The select lines are controlled by an external controller.
In another exemplary embodiment, the memory is a single port RAM that is divided into blocks based on the applicable chipping rate and the round-trip distance between serviceable mobile terminals and a base station. If the memory is divided into three blocks, for example, during each pass by a virtual or physical rake finger, two blocks in the memory are in a Read mode while the third block is in a Write mode These blocks are further divided into segments based on a performance area trade-off. In this context, each block of RAM is in effect a stand-alone single port RAM.
In another exemplary embodiment, the memory can contain enough storage locations to store data from multiple antennas and store enough data from each antenna to allow the alignment of the signals received from the multipaths.
Each block of data may contain data from a plurality of CDMA encoded channels each with possible multipaths. By reprocessing the data multiple times the memory and the rake fingers are effectively used many times per chip (or block of chips) thus significantly reducing the amount of hardware required for a given number of (logical) fingers, and combiners.
To avoid having to use a multiport read/write RAM, a time-sharing method may be used. One (or more) cycle can be dedicated to write operations and many cycles can be dedicated to read operations.
Reference will now be made in detail to the various embodiments of the invention. In the following detailed description of the present invention, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that the present invention may be practiced without these specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail so as not to unnecessarily obscure aspects of the present invention.
Data received from the front-end processor 104 is decimated by the decimation circuit 302 by a factor of D to reduce the sampling rate (R) to a minimum sampling rate (e.g., R/D) more efficient for storage in the memory 304.
Data retrieved from the memory 304 is interpolated by the interpolation circuit 306 by a factor of IR to increase its speed to a rate of R*IR/D before being outputted into the base band processor 204. In an exemplary embodiment, the decimation factor (D) and the interpolation factor (IR) are independently determined depending on the surrounding infrastructure. For example, the decimation factor may be configured to decimate the sampling rate by a factor of four while the interpolation factor may be configured to increase the sampling rate by a factor of two. When the decimation factor and the interpolation factor are the same, the decimation and the interpolation factors cancel each other exactly and the final output data rate is the same as the received data rate (i.e., sampling rate).
In one embodiment, the decimation factor and the interpolation factor are hard-coded into the decimation-interpolation circuit 202. In another embodiment, the decimation factor and the interpolation factor are dynamically configurable to adapt to multiple communication protocols by making the filter coefficients programmable.
The sequence v(n) is down sampled in the down-sampler circuit 404 by a factor D to produce y(m):
A person skilled in the art will know specific gate structures necessary to construct a decimation circuit to serve the purposes of the decimation circuit 302 based on the above described principles.
An increase in the sampling rate by an integer factor of IR can be accomplished by interpolating IR-1 new samples between successive values of the signal. As generally known in the art, the interpolation process can be accomplished in a variety of ways. In an exemplary embodiment, a process that preserves the spectral shape of the signal sequence x(n) is used.
In an exemplary embodiment, v(m) denotes a sequence that is obtained from x(n) by adding IR-1 zeros between successive values of x(n). In an exemplary embodiment, the sampling rate of v(m) is equal to the rate of y(m) above. Further, v(m) is equal to x(m/IR), when m=0, ±IR, ±2IR, . . . and equal to zero otherwise. A z-transform is performed on v(m) in the up-sampler circuit 502. Subsequent to the z-transform, the sampling rate increase obtained by the addition of IR-I zero samples between successive values of x(n) results in a signal whose spectrum V(ωy) is an IR-fold periodic repetition of the input signal spectrum X(ωx). Because only the frequency components of x(n) in the range 0≦ωy≦π/IR are unique, the images of X(ω) above ωy=π/IR are rejected by passing the sequence v(m) through the lowpass filter 504 that has a frequency response of HI(ωy). In an exemplary embodiment, HI(ωy)is equal to C when 0≦|ω|≦π/IR and equal to 0 otherwise. C is a scale factor that is required to properly normalize the output sequence y(m). In an exemplary embodiment, C is selected so that the output y(m) is equal to x(m/IR) for m=0, ±IR, ±2IR, etc. In an exemplary embodiment, the output sequence y(m) may be expressed as a convolution of the sequence v(n) (see
A person skilled in the art will know specific gate structures necessary to construct an interpolation circuit to serve the purposes of the interpolation circuit 306 based on the above described principles. Additional information regarding decimation and interpolation can be found at pages 549-555 of A. V. Oppenheim et al., Signals & Systems (Prentice Hall, 2d edition 1997). This reference is hereby incorporated for all purposes.
In an exemplary embodiment, the memory 304 stores only samples from one antenna. When multiple antennas are providing samples (e.g., at a base station), multiple memories 304 should be implemented, one per antenna.
When more than one physical rake finger is implemented in the base band processor 204, the rake finger 802 and the select circuit 602 are replicated for each physical rake finger. Each physical rake finger requires its own set of select lines 808-812. In an exemplary embodiment, rather than using a different rake finger 802 to process the data of each different multipath, circuits in the rake finger 802 can be reused. In one embodiment, if three virtual fingers are implemented in a single physical rake finger 802, the select lines 808-812 will cycle through three sets of sample select operations, one per virtual finger, so that each virtual finger is given access to all the samples retrieved from the memory 304.
In an exemplary embodiment, the memory 304 is accessible substantially simultaneously by multiple physical or virtual rake fingers 802. For example, if samples are stored in the memory 304 at an 8× over sampling rate, during access by the rake finger 802, sixteen sets of I data (or words) are received by the I multiplexers 804 and sixteen sets of Q data (or words) are received by the Q multiplexers 806. That is, for the I multiplexers 804, the same sixteen sets of I data are available to the Early, On-Time, and Late multiplexers. Similarly, for the Q multiplexers 806, the same sixteen sets of Q data are available to the Early, On-Time, and Late multiplexers.
The size of the blocks 902 is determined based on several factors, including: the chipping rate of the processing system and the maximum (serviceable) round-trip distance between a mobile terminal and a base station. The chipping rate is dependent on applicable standards (e.g., IS-95, 3GPP, etc.). Generally, the higher the chipping rate, the bigger the memory block 902 should be. Further, the longer the maximum round-trip distance, the bigger the memory block 902 should be. For example, if the maximum round-trip distance is about 10 kilometers, each block in the memory 304 should be able to store about 256 chips. A person skilled in the art knows how to calculate the memory block size based on the maximum round-trip distance and the chipping rate.
Multiple despreading circuits 600a-600b are connected to the memory segments 1002 via a bus 1010. Each despreading circuit 600 includes a selector circuit 602 and a rake finger 1008. The selector circuit 602 includes a block multiplexer 1012, three sample multiplxers 1014a-1014c, and a cache 1016. The rake finger 1008 includes hardware known in the art. The block multiplexer 1012 selects one of the bussing elements 1004 to read data from one of the segments 1002. The sample multiplexers 1014 selects the Early, On-time, and Late samples among the I and Q data pairs read from a segment 1002, which was selected by the block multiplexer 1012. The I and Q data pairs selected by the sample multiplexers 1014 are stored into the cache 1016 to be processed by the rake finger 1008. In an exemplary embodiment, the cache 1016 is a FIFO cache that begins storing samples when the first chip to be processed is presented to the FIFO. In an exemplary embodiment, each selector circuit 602 is coupled to an independent interpolation circuit (not shown) to interpolate samples retrieved from the memory 304.
The interpolation circuit 306 interpolates the rate of the data read from the memory 304 (step 1110) then outputs the data at an interpolated rate (e.g., a predetermined multiple of the chipping rate (i.e., 1×, 2×, 4×, 8×, etc.)) into the next processing module, such as the base band processor 204 (step 1112).
In an exemplary embodiment, the decimation and interpolation factors are hard-coded into the decimation-interpolation circuit 202. In another exemplary embodiment, the decimation and interpolation factors are dynamically re-programmable to accommodate various communication protocols depending on the surrounding infrastructure either by utilizing a scan chain (which is a technique known to one skilled in the art) or by accessing registers controlling the operation by the microprocessor, or by other means. Data is stored into the memory 304 at a minimum speed to reduce the required memory size. Later, data retrieved from the memory 304 is interpolated to an increased rate. As a result, memory space requirements at the memory 304 are small while efficiency during the next processing stages (such as at the base band processor 204) is not compromised.
Register 1201 is segmented into n segments, each being capable of being written independently and each storing one sample. If n is a power of 2 the operation is simpler. Successive samples are stored into successive segments in register 1201. When n samples have been assembled they are written into one memory location in memory 1202.
Each finger has its own register 1203 and multiplexer 1204.
The memory is addressed by address circuitry that comprises a plurality of multiplexers 1302-0 through 1302-7, address lines A_LOW and A_LOW+1 and select lines 0 through 7. Each multiplexer selects an address from either A_LOW or A_LOW+1 depending on the value on the select line connected to the multiplexer. The number of multiplexers (and select lines) is equal to the number of samples stored at each address 1301 in memory.
The memory address on address line A_LOW is formed by dividing the sample address by the number of samples stored at each memory address. The memory address on address line A_LOW+1 is one more than the memory address on address line A_LOW. The remainder that results from the division of the sample address by the number of samples stored controls the value on each select line. If the remainder is equal to or less than the number of the select line, the signal supplied by the select line to the multiplexer is such that the memory address on line A_LOW is selected. If the remainder is greater than the number of the select line, the signal supplied by the select line to this multiplexer is such that the memory address on line A_LOW+1 is selected.
Thus, if it is desired to read out 8 samples starting with same address 5, Select5, Select6, and Select7 will steer the multiplexer 1302 to supply A_LOW and the remaining Selects will steer their multiplexers 1302 to supply A_LOW+1. The multiplexer 1204 will supply data in the order 5, 6, 7, 8, 9, 10, 11, 12.
In the apparatus of
When a memory such as shown in
Memory 1202, if n samples wide, can support up to n−1 fingers if operated at chipping rate.
If the memory 1202 is operated at some higher frequency than the chipping rate, a multiple, plus possibly some overhead for context switching, then data from the same block of memory can be fed more than once to the same finger hardware thus providing finger reuse. A block of data, for example 256 chips, can be processed by a finger, then a second block (probably with different starting point) can be processed by the same finger hardware. As each finger is not processing a continuous data stream, cycles may be required for context switching, possibly including the initialization of interpolation filters.
If the memory 1202 is operated at some higher frequency than the chipping rate, a multiple, plus possibly some overhead for context switching, or if less than n−1 fingers are supplied, and the size of the memory is increased by a multiple A (new memory size=previous memory size×A), then data from A antennas can be supplied to the rake fingers. The memory block, with its addressable access is inherently a multiplexor.
The foregoing examples illustrate certain exemplary embodiments of the invention from which other embodiments, variations, and modifications will be apparent to those skilled in the art. The invention should therefore not be limited to the particular embodiments discussed above, but rather is defined by the claims.
This application claims priority from the following Provisional Applications: (1) “Method and Apparatus for Decimation-Interpolation Applied to CDMA Systems,” bearing U.S. Ser. No. 60/222,027, filed on Jul. 31, 2000; (2) “Apparatus and Method for Concurrent Sample Selection and Reuse of Rake Fingers,” bearing U.S. Ser. No. 60/222,008, filed on Jul. 31, 2000; (3) “Concurrent Sample Selection, Segmented-Cache for CDMA Rake Fingers,” bearing U.S. Ser. No. 60/222,025, filed on Jul. 31, 2000; and (4) “Apparatus and Method for Concurrent Sample Selection in a CDMA System,” bearing U.S. Ser. No. 60/222,029, filed on Jul. 31, 2000.
Number | Date | Country | |
---|---|---|---|
60222027 | Jul 2000 | US | |
60222008 | Jul 2000 | US | |
60222025 | Jul 2000 | US | |
60222029 | Jul 2000 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 09920095 | Jul 2001 | US |
Child | 11447228 | Jun 2006 | US |