Improved frame structure for an adaptive modulation wireless communication system

IMPROVED FRAME STRUCTURE FOR AN ADAPTIVE MODULATION WIRELESS COMMUNICATION SYSTEM

1 illustrates an exemplary cell structure having multiple CPEs and a base station associated with the cell;

2 illustrates an exemplary time division duplex (TDD) frame in accordance with the present invention;

3 is a flow diagram of an exemplary process for allocating time slots of TDD frames in accordance with the present invention;

4 is a flow diagram of an exemplary process for simplifying the structure of data inserted into a TDD frame in accordance with the present invention;

5 is a block diagram of an exemplary transmitter for use with the present invention;

6 is a block diagram of an exemplary receiver for use with the present invention;

7 is a block diagram of a prior art finite impulse response ("FIR") filter suitable for use with the present invention.

8A-8F illustrate the method of the present invention for changing the weights of an FIR filter as new symbols are passed through the filter.

<backgroundart><p>The present invention relates to a frame structure for a communication system, and more particularly, to a frame structure for an adaptive modulation wireless communication system.</p></backgroundart><abstractproblem><p>A wireless communication system facilitates two-way communication between a plurality of customer premise equipment (CPEs) and a network infrastructure. Exemplary systems include mobile cellular systems, personal cellular communications (PCS) and cordless telephones. The goal of these wireless communication systems is to provide a communication channel according to a request between a user connected to the CPE and a base station for the purpose of connecting the CPE users to a network infrastructure (usually a wire-line system). In a multiple access radio system, the basic transmission unit is usually a frame of time. Such a frame is usually subdivided into a plurality of time slots. A time slot of a frame may hold different types of data including control data and user information or data. To manage the use of a frame's time slot, the time slot may be assigned or assigned to one or more CPEs. In this case, a CPE that receives or has an assignment of a time slot may parse the assignment of a time slot between one or more users associated with the CPE. CPEs typically communicate with base stations using a "duplex" scheme that allows information to be exchanged in both directions of the connection. In this scheme, the time slot of each frame may be assigned to data transmitted from the base station to the CPE and data transmitted from the CPE to the base station.</p><p>The transmission from the base station to the CPE is commonly referred to as "downlink" transmission. The transmission from the CPE to the base station is commonly referred to as "uplink" transmission. Prior art wireless communication systems typically use Time Division Duplexing (TDD) known in the art to facilitate information exchange between a base station and a CPE. In a TDD system, duplexing of transmissions between a base station and an associated CPE is performed in the time domain. In addition, the CPE typically communicates with the associated base station on a signal having a specific predetermined radio frequency. In a TDD system, the bandwidth or channel of a signal is time-divided into frames with repeating time periods or time "slots". The time slots are used for uplink and downlink transmissions between the base station and the associated CPE.</p><p>When a wireless system runs within an area, the area is typically divided into cells with base stations located within each cell. Each base station within a cell of a wireless system ideally provides communication between CPEs located within the cell. The size or structure of a cell is generally determined as a function of the physical location of the base station, the building location, and other physical obstacles placed within the cell. The maximum bit rate per symbol modulation scheme that can be used for a cell may be limited due to channel interference within the cell and implementation of CPE or modem complexity. Channel interference can occur between adjacent time slots assigned to different CPEs in a cell due to distortion of signals between the base station and CPEs within the cell. Such signals are usually distorted by the signal's destructive multipath response, where the signal is reflected away from physical objects within the cell. In addition, signals are usually distorted by atmospheric conditions (such as rain). Thus, in order to have duplex communication between all CPEs associated with a base station in a cell, a modulation scheme having a bit rate per symbol that enables communication between all CPEs associated with a base station is selected. </p><p>However, the channel interference between the CPE and the base station varies for each CPE, for example, depending on the physical obstacle between the base station and the CPE. As a result, the maximum bit rate modulation scheme per symbol that can be used for communication between each CPE and the base station (ie, with an acceptable error rate for a given channel interference) may vary. In addition, the implementation or modem complexity of a CPE associated with a base station may vary where some CPEs may support higher bit rate per symbol modulation schemes than others associated with a base station. Thus, the selection of one low bit rate per symbol for all CPEs, where some CPEs may support higher bit rate per symbol modulation within the cell, will not maximize bandwidth utilization. The use of different or variable bit rates per symbol for different CPEs associated with a cell may improve bandwidth utilization. Unfortunately, the variable bit rate per symbol is not used for CPE-to-CPE communication associated with the base station due to its complexity. In particular, variable bit rate per symbol modulation schemes typically require complex CPE demodulators, which some CPEs have already limited implementation or modem complexity. Accordingly, there is a need for a frame structure and frame construction technique that enables variable bit rate per symbol modulation for CPEs and base stations within a cell, without increasing the complexity of the CPEs. The present invention provides such a frame structure and frame construction technique.</p></abstractproblem>

<p>The present invention includes a method for ordering or allocating downlink time slots based on the complexity of modulated data stored within the downlink time slots. Preferably, the downlink time slots are classified from the least complex modulation scheme to the most complex modulation scheme. In one embodiment, the method allocates a portion of at least two downlink time slots to at least two receiving units in which the modulation scheme used by the at least two units may vary. The method first determines the complexity of a modulation scheme used by at least two units. The method then allocates a portion of the at least two time slots to the at least two units based on the complexity of the modulation scheme they use. As noted, ideally, some of the at least two downlink time slots are allocated from the least complex modulation scheme to the most complex modulation scheme. In another embodiment, the method may first order at least two units as a function of the complexity of the modulation scheme they use. The method may then allocate a portion of the at least two time slots based on the order of the at least two units.</p><p>Further, the method may order the uplink time slots of the frame based on the complexity of the modulated data to be stored in the uplink time slots. Preferably, the uplink time slots are classified from the least complex modulation scheme to the most complex modulation scheme. In one embodiment, the method allocates at least two uplink time slots to at least two transmitting units in which the modulation scheme used by the at least two transmitting units may vary. The method first determines the complexity of a modulation scheme used by at least two transmitting units. The method then allocates at least two time slots to the at least two transmission units based on the complexity of the modulation scheme they use. As noted, ideally the at least two uplink time slots are allocated from the least complex modulation scheme to the most complex modulation scheme. In another embodiment, the method may first order at least two transmission units as a function of the complexity of the modulation scheme they use. The method may then allocate at least two uplink time slots based on the order of the at least two transmission units.</p><p>The present invention also includes a method of ordering downlink time slots based on a bit rate per symbol of a modulation scheme used to generate data stored within the downlink time slots. Advantageously, said downlink time slots are classified from a lowest bit rate per symbol modulation scheme to a highest bit rate per symbol modulation scheme. In one embodiment, the method allocates a portion of at least two downlink time slots to at least two receiving units in which the bit rate per symbol modulation scheme used by the at least two units may vary. The method first determines a bit rate per symbol of a modulation scheme used by at least two units. The method then allocates a portion of the at least two time slots to the at least two units based on the bit rate per symbol modulation scheme they use. As noted, ideally a portion of at least two downlink time slots is allocated from the lowest bit rate per symbol to the highest bit rate per symbol. In another embodiment, the method may first order a portion of the at least two units according to the bit rate per symbol of the modulation scheme they use. The method may then allocate a portion of the at least two time slots based on the order of the at least two units.</p><p>The method may also order the uplink time slots of the frame based on the bit rate per symbol of the modulation scheme used to generate the data stored within the uplink time slots. Preferably the uplink time slots are also classified from the lowest bit rate per symbol to the highest bit rate per symbol. In one embodiment, the method allocates at least two uplink time slots to at least two transmitting units in which the bit rate per symbol of the modulation scheme used by the at least two transmitting units may vary. The method first determines a bit rate per symbol of a modulation scheme used by at least two transmitting units. The method then allocates at least two time slots to the at least two transmission units based on the bit rate per symbol of the modulation scheme they use. As noted, ideally at least two uplink time slots are allocated from the lowest bit rate per symbol modulation scheme to the highest bit rate per symbol modulation scheme. In another embodiment, the method may first order at least two transmission units as a function of the bit rate per symbol modulation scheme they use. The method may then allocate at least two uplink time slots based on the order of the at least two transmission units.</p><p>The invention also includes a method for determining to encode an Ld bit of data into a frame. A frame has a time length T, and the frame is transmitted at a baud rate R. The method first determines a maximum fixed per symbol bit rate of modulation for the Ld bits of data. Then the method is<i> (R*T*</i><i>Bi</i><i>)/(</i><i>Ld</i><i>+x)</i>Adds x error code bits so that is an integer, where Bi is the bit rate per symbol of the modulation scheme used. Note that x may have a minimum value based on the minimum block error rate. Also, x error code bits may be Reed-Solomon encoded error bits. In another embodiment, the method may determine the maximum bit rate per symbol, Bi of the modulation scheme for Ld bits of data. Then the method is<i>(R*T*</i><i>Bi</i><i>)/(</i><i>Ld</i><i>+x)</i>You can add x error code bits so that is an integer. </p><p>In another embodiment, the method first selects a convolution ratio, wherein the selected convolution ratio adds y convolution bits to the Ld bits of the data after convolutional encoding of the Ld bits of the data. Then the method is<i>(R*T*</i><i>Bi</i><i>)/(</i><i>Ld</i><i>+x+y)</i>Add x error code bits so that is an integer. In this method, the convolution ratio is<i>(R*T*</i><i>Bi</i><i>)/(</i><i>Ld</i><i>+x+y)</i>may be modified to be an integer. In addition, the number of x error bits is<i>(R*T*</i><i>Bi</i><i>)/(</i><i>Ld</i><i>+x+y)</i>may be chosen to be an integer. </p><p>Further, the present invention includes a method for determining a modulation scheme of a frame having a plurality of downlink time slots, one of the plurality of downlink time slots including control information. In this method, the modulation scheme used to generate modulated data for the plurality of downlink time slots may be changed for each of the plurality of downlink slots. In addition, the frame may be transmitted in a plurality of units, each of which can support a modulation scheme having a maximum complexity. The method first determines the lowest modem complexity supported by each of the plurality of units. Then, the method sets the modulation complexity of the downlink slot of the plurality of downlink time slots including the control information for the determined lowest modem complexity.</p><p>In this method, the downlink slot of the plurality of downlink slots including the control information may be a first downlink slot within a time sequence of the plurality of downlink slots. In addition, the method may also determine a complexity of a modulation scheme used to generate modulated data for at least two units of the plurality of units. The method may then allocate at least two of the plurality of time slots to the at least two units based on a complexity of a modulation scheme used to generate modulated data for the at least two units. The allocation for at least two units may be from the least complex modulation scheme to the most complex modulation scheme.</p><p>The invention also includes a method for setting a weight value of a finite impulse response filter. In this case, the filter receives a symbol with a variable modulation rate and stores a plurality of symbols, each stored symbol having a corresponding weight. The method first determines when a first symbol is received with a different modulation rate than the last stored symbol. The method then changes the weight value corresponding to the first symbol based on the modulation rate of the first symbol. The method may also include receiving a second symbol having a modulation rate equal to a modulation rate of the first symbol. Thereafter, the weight value corresponding to the first symbol is changed based on the modulation rate of the first symbol. More generally, the method changes the weight value corresponding to the first symbol based on the modulation rate of the first symbol as it passes through the filter.</p><p>Preferred and other embodiments of the present invention are described in the accompanying drawings and detailed description. Numerous additional modifications and variations will become apparent to those skilled in the art once the detailed description of the invention is known.</p><p>The preferred embodiments and examples described herein are to be regarded as illustrative and not restrictive of the present invention. </p><p>The present invention includes an improved frame structure and a process for creating the frame structure for use in a wireless communication system that utilizes adaptive modulation. Adaptive modulation involves varying the per-symbol bit rate modulation scheme or modem complexity of the signal transmitted between the CPE and the base station as a function of the channel interference of the signal or the implementation or modem complexity of the CPE. 1 is a diagram of an exemplary cell 10 comprising a centrally located base station 20 within the cell 10 and a plurality of CPEs 30 , 32 , 34 , 36 , 38 associated with the base station. 1 does not show buildings or other physical obstacles (eg trees or hills) that may cause channel interference between the signals of the CPE.</p><p>As noted above, the maximum bit rate per symbol modulation scheme or technique or the most complex modulation scheme selected for use in cell 10 is typically determined as a function of the channel interference between the CPE and the CPE implementation or modem complexity. Also, as noted above, selection of a single maximum bit rate per symbol modulation technique based on the lowest bit rate per symbol modulation scheme supported by all CPEs may not optimize bandwidth utilization within the cell 10 . In particular, lower channel interference between some CPEs (such as units 38 and 30) may allow the use of higher bit modulation techniques or more complex modulation schemes with error levels below the maximum desired error level. have. However, adaptive bit rate modulation or variable bit rate modulation between different CPEs usually requires complex transceivers within the CPE so that the CPE can have limited implementation or modem complexity ahead of time.</p><p>As is well known, the frame structure is divided into a plurality of downlink and uplink slots. Each downlink time slot can be used to store data received by multiple users whose data the user recognizes by one address or another label. Uplink time slots are typically assigned to individual users for data transmission from a user to another user or system via a base station. To maximize bandwidth utilization and minimize modulator complexity in the base station and associated CPE, the present invention simplifies the structure of data inserted into a time slot. Briefly, data blocks are ideally parsed into integers of time slots. These processes will be described below with reference to FIG. 4 . Second, the present invention orders or classifies the placement of data within the downlink and uplink time slots as a function of the modem complexity or bit rate per symbol modulation scheme provided for generating the data placed within the time slots. 3, this technique reduces the complexity of the CPE modulator and the number of modulation scheme variations within a frame. </p><p>Figure 2 shows that an adaptive bit rate per symbol modulation scheme is used within a cell that enables it to be used within a frame structure without increasing the complexity of the receiver and transmitter of the CPE associated with the cell or reducing the number of modulation scheme variations within each frame. A diagram illustrating an exemplary frame structure. As shown in FIG. 2 , a frame 80 includes a plurality of time slots. In this example, the first five time slots contain downlink data 82 (from the base station 10), and the remaining five time slots (from the CPE to the base station 10) ) there are 10 time slots containing uplink data 84 . In this example, the downlink slot is DM<sb>1</sb>, DM<sb>2</sb>, DM<sb>3</sb> and DM<sb>4</sb>A modulation bit rate per symbol, wherein the four downlink time slots are assigned to at least four CPEs, and the CPEs will reclaim data located within these slots based on their respective assignments. Multiple CPEs may be assigned to any one downlink time slot, where each CPE retrieves its data from this slot based on an address or identifier. As a result, the CPE can only retrieve data from a part of the downlink time slot.</p><p>In addition, the uplink slot is UM<sb>1</sb>, UM<sb>2</sb>, UM<sb>3</sb> and UM<sb>4</sb>A modulation bit rate per symbol, where the four uplink time slots are typically assigned to four CPEs, and the CPEs will insert data into these slots based on their respective assignments. Note that in some embodiments, a CPE may be assigned to more than one uplink slot. In addition, the downlink control information may be located at the start time of the downlink time slot, and the unreserved time slot may be located at the start time of the uplink time slot. It is clearly desirable that any CPE associated with a cell can retrieve data located within the downlink control information time slot regardless of the CPE's location within the cell. In addition, each CPE must be able to insert data into unreserved uplink time slots.</p><p>As described above, in an adaptive bit rate per symbol modulation system, the modulation scheme may vary for each CPE and thus for each downlink and uplink time slot. In order to minimize the complexity of the CPE and the base station used in the system, and to reduce the number of modulation scheme variations in a frame, the present invention provides a DM<sb>1</sb>DM<sb>2</sb>DM<sb>3</sb>DM<sb>4</sb> and UM<sb>1</sb>UM<sb>2</sb>UM<sb>3</sb>UM<sb>4</sb>need to work Thus, ideally, the data within a time slot is arranged from the least complex modulation scheme to the most complex modulation scheme. As noted, this technique reduces the number of modulation variations, which can simplify the implementation of a base station using the frame structure 80 . It also allows the base station and the CPE to train on the least complex data, which can reduce the error rate.</p><p>Also, ideally the downlink control information is encoded using the least complex modulation scheme of the system, and information placed within unreserved uplink time slots is also ideally encoded using the least complex modulation scheme of the system. . This ensures that all CPEs associated with a cell will be able to receive or encode information within a desired level of error. Ideally, the control information indicates where within the frame the modulation transition occurs. An exemplary process 90 of allocating time slots of the frame 80 shown in FIG. 2 is shown with reference to FIG. 3 .</p><p>As shown in Figure 3, a first step 92 of process 90 includes determining which CPE will receive in at least one time slot within a next frame. In a duplex system as described above, a CPE that receives data in a downlink time slot may transmit data in an uplink time slot. In other systems, such as point-to-multipoint or multicast systems, there may be more downlink time slots than uplink time slots. Then (at step 94) the maximum complex modulation scheme or maximum bit rate per symbol of the modulation scheme used by the CPE is determined for each CPE. As described above, the maximum complex modulation scheme or maximum bit rate per symbol modulation scheme can be determined as a function of the channel interference and maximum desired error level of the signal of the CPE and the implementation or modem complexity of the CPE.</p><p>In a preferred embodiment, the binary phase shift keying scheme (hereinafter referred to as 'BPSK') modulation may be selected for the least complex modulation scheme. In BPSK, the bit rate Bi per symbol of the modulation scheme is 1. That is, each symbol represents one bit. Bi may also be referred to as modulation scheme efficiency. That is, how efficiently the system encodes the data. 4 Quadrature Amplitude Modulation (QAM) may be used for the intermediate modulation scheme. For QAM 4, the bit rate Bi per symbol of the modulation scheme is 2. That is, each symbol represents 2 bits. Larger quadrature amplitude modulation can be used for more complex modulation schemes, for example, QAM 64, where the modulation scheme has a bit rate Bi of 6 per symbol, each symbol represents 6 bits. The modulation complexity or bit rate per symbol modulation scheme can be modified from frame to frame or can be kept constant for multiple frames for a particular CPE. In addition, the CPE may select or indicate a desired modulation complexity or scheme. </p><p>When determining for each of the CPEs the modulation complexity or bit rate per symbol modulation scheme used to encode the data, in step 96, the CPEs are sorted in ascending order based on the selected modulation complexity or bit rate per symbol modulation scheme. That is, it is classified from the lowest bit rate per symbol modulation scheme to the highest bit rate per symbol modulation scheme, or from the least complex modulation scheme to the most complex modulation scheme. Finally, the time slots of a frame are placed or assigned to the CPEs in a classification order from the lowest bit rate per symbol modulation scheme to the highest bit rate per symbol modulation scheme, or from the least complex modulation scheme to the most complex modulation scheme. As discussed above, frames are constructed using this process, thereby reducing the complexity of the base station and CPE inserting or retrieving data therefrom. Even if the modulation scheme varies from CPE to CPE, the number of symbols transmitted in burst is a predetermined number for all CPEs regardless of the modulation scheme. <i>nxS</i>Note that it is usually fixed as </p><p>group of symbols <i>nxS</i>It is desirable to simplify the configuration of the time slot given the fixed burst and variable modulation scheme of Modulation of L bits produces a fixed number of symbols S, where S=(L|B<sb>1</sb>) and B<sb>1</sb>Note that is the bit rate per symbol of the modulation scheme. To simplify time slot usage and bandwidth management, (L|B<sb>1</sb>) or S is ideally an integer multiple of the product of the length of the time slot Ts and the baud rate R of the frame. Thus, ideally the L bits fit into an integer number of time slots Ts based on the modulation scheme. Each frame has a fixed number of time slots, where the length of the frame (i.e. the number of time slots) is the maximum desired delay T between the system's signal transmission and the baud rate R (symbols transmitted per second).<sb>D</sb>Note that it determines the function of Thus, for each frame, the number of transmitted symbols is T<sb>D</sb>Equivalent to *R. number of symbols<i>nxS</i> or (L|B<sb>1</sb>) is preferably an integer multiple of the number of symbols transmitted per frame. Therefore, the ratio (T<sb>D</sb>*R)(LB<sb>1</sb>) is preferably an integer. rain (T<sb>D</sb>*R)(LB<sb>1</sb>) is an integer, <i>nxS</i><i></i>A fixed number of bursts of symbols may be transmitted within each frame. This can simplify frame usage and bandwidth management.</p><p>In most systems, L bits of data represent an encoded signal containing overhead or Forward Error Correction (FEC) information, where L bits of L<sb>D</sb>Only the pure data to be transmitted to the unit or base station. In these systems, data bits L that are transmitted in bursts<sb>D</sb>The number of may be fixed to, for example, 256, 512, or 1024 bits. FEC information typically includes a block code containing convolutional encoding bits and error correction encoding bits such as Reed-Solomon (RS(n,k)) data. In another embodiment, convolutionally encoded data may also be interleaved prior to error encoding. T<sb>D</sb>,R and S are fixed due to system constraint, B<sb>1</sb>L is selected as a function of channel interference and modem or implementation complexity, then L is ideally modified to simplify time slot configuration or bandwidth management of frames. As noted, L<sb>D</sb> It can also be fixed within this system. In such a system L will be determined for each possible modulation scheme of the system. 4 shows T for transmission of data by a unit or base station so that frame usage is simplified;<sb>D</sb>,R and B<sb>1</sb>A flow diagram of a preferred process 60 for constructing or determining L based on </p><p>As shown in Figure 4, the first step 62 of the process 60 is the maximum allowable delay T of the system.<sb>D</sb>to decide As mentioned above, delay T<sb>D</sb>is set equal to the maximum allowable delay between the transmission of signals between the CPE or unit and the base station. In step 64, L<sb>D</sb> The maximum bit rate per symbol modulation scheme or the most complex modulation scheme that can be used for transmission of bits is determined or selected (the process is as described above). Then in step 66, the convolution ratio (<i>x/y</i>) is L<sb>D</sb> selected for the data bits. In some embodiments, no convolutional encoding is used. In this example, (<i>x/y</i>) is set to 1. convolution ratio (<i>x/y</i>) is the data L<sb>D</sb> It is one parameter that can be modified to change the number of bits needed to encode the bits. In step 68, an error encoding level is selected as another variable parameter. The block code is the L of the data<sb>D</sb> It is used to reduce the bit block error rate (hereinafter referred to as 'BER') to a desired level. In a preferred embodiment, a Reed-Solomon (RS) block code is used. Therefore, the L of the data<sb>D</sb> The number of bits L required to encode the bits depends on the convolution ratio (<i>x/y</i>) and the selection of the error code level. </p><p>In step 72, <i>(T</i><i><sb>D</sb></i><i>*R)(</i><i>L1B</i><i><sb>1</sb></i><i>)</i>The value of the ratio Z is determined. Baud rate R is fixed, delay T<sb>D</sb>is determined in step 62, and B<sb>1</sb>is determined in step 64, and L is determined as a function of the parameter selected in steps 66 and 68. When it is determined in step 74 that the ratio Z is not an integer, a different convolution ratio (in step 66) or an error code level (in step 68) may be selected. In a preferred embodiment, the selection of the convolution ratio and the error code level is varied as a function of the fraction of the ratio Z. That is, a convergence algorithm may be used. As noted above, in some embodiments, the convolution ratio is fixed at 1. In such an embodiment, only the error code or block code level is corrected. To ensure that the ratio Z is an integer, the number of bits used to generate the block code of data may be greater than that required to meet the minimum BER. In step 74, the ratio Z is determined to be an integer, then the process is complete and the block of L bits is optimized or simplified for the modulation scheme or bit rate Bi per symbol.</p><p>A transmitter 40 and a receiver 50 that can be used to transmit and receive frames of data according to the present invention will be described with reference to FIGS. 5 and 6 . 5 is a block diagram of an exemplary transmitter 40 . As shown in the figure, the transmitter 40 includes a convolutional encoder 42 , a block encoder 44 , an M-ary modulator 46 , a frame generator 48 , and an up-converter 49 . Transmitter 40 L of data<sb>D</sb> It receives bits, encodes the data to produce L bits of data, packs the L bits of data into frames, and up-converts the frames of data to a transmission frequency. Convolutional encoder 42 and block encoder 44<sb>D</sb> Provides FEC data that converts bits into L bits of data. In particular, the convolutional encoder 42<sb>D</sb> The ratio (<i>x/y</i>) is used. The block encoder uses the selected code level to encode the convolutional data to generate L bits of encoded data that is sent to the base station or unit.</p><p>The M-ary modulator then converts the L bits of data based on the selected bit rate per symbol Bi <i>nxS</i> convert to symbol. Due to the choice of convolution ratio and error code level,<i>nxS</i> A symbol may be inserted as an integer number of time slots in a frame. Frame generator 48 would ideally insert nxS symbols into the time slots of the frame based on the process shown with reference to FIG. 3 above, i.e., in the order of the modulation schemes (least complexity to highest complexity modulation scheme). The up-converter 49 frequency-shifts a frame of data bundled with a frequency suitable for transmission between a CPE or a unit and a base station based on a technique known to those skilled in the art.</p><p>The receiver 50 shown in FIG. 6 converts the frame of frequency-shifted data back to L of data.<sb>D</sb> Switch to bit group. As shown in FIG. 6 , the receiver 50 includes a down converter 59 , a frame destructor 58 , an M-ary modulator 56 , a block decoder 54 , and a convolutional decoder 52 . The down converter 59 frequency shifts the received signal to baseband using techniques known in the art. The frame destructor is for processing by the remaining components of the receiver 50.<i>nxS</i> Separate frames into groups of symbols. When the receiver 50 is part of a subscriber unit, the frame destructor ensures that data is directed to the subscriber unit.<i>nxS</i> Select one of the groups of symbols. The block decoder 54 may be configured using techniques known to those skilled in the art.<i>nxS</i> Decode the symbol. The convolutional decoder then decodes the data to<sb>D</sb> create a bit </p><p>The techniques and systems described above may be modified within the scope of the appended claims. For example, the symbol shape may also be used in the preferred embodiment to avoid spectral leakage due to possible abrupt changes in the modulation scheme of the frame as described above. The symbol shape is passed through the FIR filter as an exemplary prior art FIR filter 60 is shown in FIG. 7 .<i>nxS</i> This is usually accomplished by filtering the symbols. As shown in FIG. 7 , the FIR filter 60 includes a k multiplier and a sum node 66 . Symbol S is sequentially received and stored in filter taps TO through Tk (62). Each multiplier 64 has tap weights W0 through Wk and taps T0 through Tk associated with the symbol stored in tab 62 . As can be seen from Figure 7, the FIR filter 60 is<img file="KR100689175B1_D0001.tif" /> Produces an output y of the form</p><p>Other modulation schemes, such as other QAM schemes (QAM-4, QAM-16, QAM-64) <i>x </i>Note that we use different "alphabet" to represent symbols. For example, QAM-4 has 4 different symbols, QAM-16 has 16 different symbols, and QAM-64 has 64 different symbols. In addition, different modulation schemes may have different gains applied to symbols for transmission due to varying back-off requirements. In prior art variable modulation systems, the memory of the FIR filter is normally reset when the modulation scheme is changed, and the weights W0 to Wk are immediately changed to weights optimized for the modulation scheme or preference of the scheme to prevent spectral leakage.</p><p>However, this solution is not ideal since the weights are not optimized for the symbols in the memory (tap 62) of the filter corresponding to the previous modulation scheme or rate. One solution is to use one set of weights for all modulation schemes. However, this solution is also not ideal since the FIR filter is not optimized with the alphabet of symbols for each modulation scheme. In order to prevent spectral leakage and optimize the FIR filter 60, the present invention sequentially changes the filter taps from the new modulation scheme to each new symbol as shown in Figs. 8A-8F. In particular, the weight corresponding to the first new symbol of the new modulation scheme is modified as the first new symbol is passed through the filter 60 . In Fig. 8a, the filter weights W0 to Wk are optimized for the modulation scheme of the symbol currently being processed by the FIR filter 60. As shown in Figs. In FIG. 8b , the first symbol of the new modulation scheme is received in T0 of the FIR filter 60 . At this point, as shown in Figure 8b, the present invention replaces the filter weight W0 associated with T0 with a new filter weight WO', where WO' is optimized for the modulation scheme of the first new symbol stored in T0. Then, when the next symbol from the new modulation scheme is received and stored in T0, and the first new symbol moves to T1, the present invention replaces the filter weight W1 associated with T1 with the new filter weight W1', where W1' is also optimized for the new modulation scheme as shown in Fig. 8c. This process is as shown in Figures 8d-8f, where all symbols are stored in tap 62, FIR filter 60 belongs to a new modulation scheme, and all filter weights WO to Wk are associated with the new modulation scheme or Repeat until optimized. This technique optimizes the weights used in the FIR filter 60 to shape the symbol or to vary the modulation scheme to reduce spectral leakage. </p><p>Accordingly, the present invention is not intended to be limited to the specific embodiments, but is intended to be limited only by the scope of the appended claims.</p>

<p>The present invention provides a method for determining the weight of a finite impulse filter.</p>