International Journal of Computational Engineering Research(IJCER)

International Journal Of Computational Engineering Research (ijceronline.com) Vol. 2 Issue. 8

Implementation of an OFDM FFT Kernel for WiMAX
Lokesh C. 1, Dr. Nataraj K. R.2.
1
Assistant Professor,Department of Electrical and Electronics Engineering.Vidyavardhaka College of Engineering,
Mysore, Karnataka, India.
2
Professor, Department of Electronics and Communications Engineering.SJB Institute of Technology, Bangalore,
Karnataka, India.

Abstract:
In this paper we focus on the OFDM Kernel which refers to the inverse fast Fourier transform and cyclic
prefix insertion blocks in the downlink flow and the FFT and cyclic prefix removal blocks in the uplink flow. To
support orthogonal frequency-division multiple access (OFDMA) an extension to the OFDM kernel is required that
allows each user to be allocated with a portion of the available carriers. This process is referred to as sub
channelization. The WiMAX building blocks include bit-level, OFDMA symbol-level, and digital intermediate
frequency processing blocks. For bit-level processing, Altera provides symbol mapping/demapping reference
designs and support for forward error correction using the Reed- Solomon and Viterbi MegaCore® functions. The
OFDMA symbol-level processing blocks include reference designs that demonstrate subchannelization and
desubchannelization with cyclic prefix insertion supported by the fast Fourier transform, and inverse fast Fourier
transform MegaCore functions. Other OFDMA symbol-level reference designs illustrate ranging, channel
estimation, and channel equalization. The digital IF processing blocks include single antenna and multiantenna
digital up converter and digital down converter reference designs, and advanced crest-factor reduction and digital
predistortion.

Keywords: inverse fast Fourier transform (IFFT), orthogonal frequency-division multiple access (OFDMA),
intermediate frequency (IF), forward error correction (FEC), digital up converter (DUC), digital down converter
(DDC), crest-factor reduction (CFR), digital predistortion (DPD, WiMAX (Worldwide Interoperability for
Microwave Access).

1. Introduction
The Altera® orthogonal frequency division multiplexing (OFDM) kernel can be used to accelerate the
development of wireless OFDM transceivers such as those required for the deployment of mobile broadband
wireless networks based on the IEEE 802.16 standard. OFDM is one of the key physical layer components
associated with mobile worldwide interoperability for microwave access (WiMAX) and is widely regarded as an
enabling technology for future broadband wireless protocols including the 3GPP and 3GPP2 long term evolution
standards.

The OFDM kernel has the following key features:

 Support for 128, 512, 1K, and 2K FFT sizes to address variable bandwidths from 1.25 to 20 MHz
 Parameterizable design
 Optimized for efficient use of Cyclone II, Stratix II, and Stratix III device resources

2. Introduction to WiMAX
WiMAX (Worldwide Interoperability for Microwave Access) is a wireless communications standard
designed to provide 30 to 40 megabit-per-second data rates, with the 2011 update providing up to 1 Gbit/s for fixed
stations. WiMAX refers to interoperable implementations of the IEEE 802.16 family of wireless-networks standards
ratified by the WiMAX Forum. Similarly, Wi-Fi, refers to interoperable implementations of the IEEE 802.11
Wireless LAN standards certified by the Wi-Fi Alliance. WiMAX Forum certification allows vendors to sell fixed or
mobile products as WiMAX certified, thus ensuring a level of interoperability with other certified products, as long
as they fit the same profile.

Issn 2250-3005(online) December| 2012 Page 74


MAC/PHY Interface

Randomization Derandomization
Uplink
Downlink
FEC Encoding FEC Decoding

Interleaving Deinterleaving
Bit Level Processing

Symbol Mapping Symbol Demapping

Subchannalization Channel Estimation and Equilization To MAC
Plot Insertion

Desubchannalization
plot extension OFDMA Ranging
IFFT

FFT
OFDM Symbol Level
Cyclic Prefix
Processing
Remove Cyclic prefix

DUC DDC

CFR Digital IF Processing
From ADC

DPD

To DAC

Figure 1. WiMAX Physical Layer Implementatio, an overview of the IEEE 802.16e-2005 scalable orthogonal
frequency-division multiple access (OFDMA) physical layer (PHY) for WiMAX basestations.

Altera’s WiMAX building blocks include bit-level, OFDMA symbol-level, and digital intermediate
frequency (IF) processing blocks. For bit-level processing, Altera provides symbol mapping/demapping reference
designs and support for forward error correction using the Reed- Solomon and Viterbi MegaCore® functions. The
OFDMA symbol-level processing blocks include reference designs that demonstrate subchannelization and
desubchannelization with cyclic prefix insertion supported by the fast Fourier transform, and inverse fast Fourier
transform MegaCore functions. Other OFDMA symbol-level reference designs illustrate ranging, channel
estimation, and channel equalization. The digital IF processing blocks include single antenna and multiantenna
digital up converter and digital down converter reference designs, and advanced crest-factor reduction and digital
predistortion.

3. Introduction to OFDM Kernal
The OFDM Kernel refers to the inverse fast Fourier transform (IFFT) and cyclic prefix insertion blocks in
the downlink flow and the FFT and cyclic prefix removal blocks in the uplink flow. To support orthogonal
frequency-division multiple access (OFDMA) an extension to the OFDM kernel is required that allows each user to
be allocated with a portion of the available carriers. This process is referred to as subchannelization. The physical
layer is based around OFDM modulation. Data is mapped in the frequency domain onto the available carriers. For
this data to be conveyed across a radio channel, it is transformed into the time domain using an inverse fast Fourier
transform (IFFT) operation. To provide multipath immunity and tolerance for synchronization errors, a cyclic prefix
is added to the time domain representation of the data. Multiple modes are supported to accommodate variable
channel bandwidths. This scalable architecture is achieved by using different FFT/IFFT sizes. This reference design
supports transform sizes of 128, 512, 1,024, and 2,048.


4. Implementing OFDM Kernal for WiMAX
FPGAs are well suited to FFT and IFFT processing because they are capable of high speed complex
multiplications. DSP devices typically have up to eight dedicated multipliers, whereas the Stratix III EP3SE110
FPGA has 112 DSP blocks that offer a throughput of nearly 500 GMACs and can support up to 896 18x18
multipliers, which is an order of magnitude higher than current DSP devices.
Sign a
Zero Loopback sign b
Zero Acc round
clk [3..0] zero Chain ouyt saturate
ena [3..0] Round Chain out rotate
Chain In Chain out
alcr [3..0] Saturate Chain out Shift Saturation /
Previous Cascade Overflow

Data A0
First Stage Adder

Second Stage Adder / Accumulator

Second Adder Register Bank

Second Round / Saturation
Data B0
First Round / Saturate
Pipelined Register bank

Output Register bank
Input register bank

Chain out Adder
Data A1

Shift / Rotate
Data Out

MUX
Data B1

Data A2
First Stage Adder

Data B2
Data A3

Data B3

Half DSP Block

Next Cascade

Figure2. Embedded DSP Blocks Architecture in Stratix III Devices

Such a massive difference in signal processing capability between FPGAs and DSP devices is further
accentuated when dealing with basestations that employ advanced, multiple antenna techniques such as space time
codes (STC), beam forming, and multiple-input multiple-output (MIMO) schemes. The combination of OFDMA
and MIMO is widely regarded as a key enabler of higher data rates in current and future WiMAX and 3GPP long
term evolution (LTE) wireless systems. When multiple transmit and receive antennas are employed at a basestation,
the OFDMA symbol processing functions have to be implemented for each antenna stream separately before MIMO
decoding is performed. The symbol-level complexity grows linearly with the number of antennas implemented on
DSPs that perform serial operations. For example, for two transmit and two receive antennas the FFT and IFFT
functions for WiMAX take up approximately 60% of a 1-GHz DSP core when the transform size is 2,048 points. In
contrast, a multiple antenna-based implementation scales very efficiently when implemented with FPGAs. Using
Altera devices, we can exploit parallel processing and time-multiplexing between the data from multiple antennas.
The same 2×2 antenna FFT/IFFT configuration uses less than 10% of a Stratix II 2S60 device.


5. Functional description of OFDM kernal for WiMAX
Altera provides the reference design as clear text VHDL. The reference design also demonstrates the use of
the FFT MegaCore function. To accelerate integration with Altera intellectual property (IP) or other subsystems, the
interfaces support the Altera Avalon® Streaming (Avalon-ST) interface specification. Altera has verified the RTL
behavior against a fixed point model of the algorithms. The reference design includes RTL testbenches that
stimulate the designs using golden reference data from the fixed point model and check for correct functionality.
The OFDM kernel handles the FFT operations and cyclic prefix addition and removal. The FFT size is a parameter
that we must specify at synthesis time, but we can change the guard interval at run time.

Downlink Transmit
The downlink OFDM kernel module performs an inverse Fourier transform of the frequency domain input
data and adds a cyclic prefix to the resulting time domain data. The cyclic prefix addition block contains a controller
that buffers the output packets from the FFT, and adds the appropriate proportion of the end of the output packet to
the beginning of the output packet. As this requires a fairly significant memory resource, the hardware architecture
has been designed so that the embedded memory may be shared with the uplink OFDM kernel if the modem is
operating in time division duplex (TDD) mode.

Avalon – ST Embedded
status source memory
Interface

Avalon – ST
Avalon – ST IFFT Mega Add Cyclic data source
Data Sink Core Function Prefix Interface
Interface

Clock 2
Clock 1
Figure 3 shows a block diagram of the downlink OFDM kernel.

Interface Specifications
The block has two clock domains. In addition, there are two reset ports; one for each clock domain. The
reset ports are active low. Figure 4 shows the downlink OFDM kernel interfaces.

cp_width Configuration Interface cp_width_out

din_ready dout_ready
cp_width
din_valid dout_valid
Avalon – Avalon –
cp_width
din_real ST Sink ST Source
Interface dout_r
Interface
din_imag cp_width
dout_i

din_start_block
dout_exp

add_cp_rdy Status source Interface dout_start_block

Figure4. Downlink OFDM Kernel Interfaces



The input interface has the following features:

 Avalon-ST data sink and status source
 Ready signal latency of one cycle—the earliest time valid data may be presented to the block after ready has
been signaled is one clock cycle

The output interface has the following features:

 Avalon-ST data source
 Ready signal latency of four cycles—the block responds to new data or stops delivering data four cycles after
an event on the ready signal
 Support for back pressure and Dynamically changeable cyclic prefix

Uplink Receive
The uplink OFDM kernel module performs an FFT of the time domain input data and removes the cyclic
prefix. The Avalon-ST start of packet pulse should specify the start of the cyclic prefix. The remove cyclic prefix
block ignores the data during the cyclic prefix and writes the remaining samples to the FFT input buffer.

Embedded
memory

Avalon – ST IFFT Mega Add Cyclic Avalon – ST
Data Sink Core Function Prefix data source
Interface Interface

Clock 1 Clock 2
Figure5. Uplink OFDM Kernel Block Diagram

Because the channel characteristics can change, it is possible that the start of the packet pulse is not always
after the start of the cyclic prefix time. The hardware has been designed to deal with this scenario but with the
constraints that the variation of the pulse must be within the cyclic prefix time and that the start pulse will not be
before the preceding symbol has been fully clocked in.

Uplink Interface Specifications
The block has two clock domains. In addition, there are two reset ports; one for each clock domain. The
reset ports are active low.

Configuration Interface dout_ready
cp_width
cp_width
dout_valid
din_ready
cp_width
dout_r
din_valid
Avalon – Avalon – cp_width
dout_i
din_real ST Sink ST Source
Interface Interface
din_imag dout_exp

din_start_block dout_start_block

dout_end_block

Figure6. Uplink OFDM Kernel Interfaces


The input interface has the following features:

 Avalon-ST data sink
 Ready signal latency of one cycle
 Does not apply back pressure since data is continuous from RF card

The output interface has the following features:

 Avalon-ST data source
 Ready signal latency of one cycle
 Does not accept back pressure from downstream sink
 Dynamically changeable cyclic prefix

FFT MegaCore Function
The FFT MegaCore function is capable of performing both the forward and inverse transform. The
hardware architecture is chosen to minimize the resource usage and has the following parameters:

 Burst mode
 Single output engine
 Single instance of engine
 16-bit internal and data input/output precision widths

In addition, design implements two clock domains so that it is possible to exploit time sharing and
minimize resource utilization in the FFT MegaCore function by running Clock 1 faster than Clock 2. The FFT
MegaCore function generates block floating point output data and the output dynamic range is maximized for the
given input and output data widths.

Clock Requirements
The clocking requirements are as follows:

 The two clock domains must be synchronous
 The minimum Clock 2 frequency is the data sampling frequency given in Table 1. This would lead to a
constant output from the FFT MegaCore function

FFT Points Bandwidth (MHz) Clock 2 (MHz)
128 1.25 1.429
256 2.5 2.857
512 5 5.714
1,024 10 11.429
2,048 20 22.857

Table 1. Minimum Clock 2 Rate
 The Clock 2 frequency may equal or exceed the Clock 1 frequency
 The Clock 1 requirements are dictated by the FFT MegaCore function and are summarized in Table 2

FFT Points Required data rate FFT throughput Clock 1 minimum
(MHz) (Cycles/ N block) Speed (MHz)
128 1.429 858 9.579
256 2.857 1,626 18.146
512 5.714 3,693 41.214
1,024 11.429 7,277 81.220
2,048 22.857 16,512 184.285
Table 2. Clock 1 Requirements



 Table 2 ignores the cyclic prefix effect, which reduces the Clock 1 speed requirement slightly
 The Clock 1 minimum speed = throughput/N × data rate

6. Conclusion
This application note has outlined the advantages of using Altera FPGAs for implementing OFDM systems
such as an IEEE 802.16e deployment. A flexible, high-throughput DSP platform needs an FPGA-based
implementation platform. In addition, this reference design demonstrates the implementation of a key function that
may be exploited to facilitate rapid system deployment.

7. Results

Symbol generator simulation output of an OFDM kernel for WiMAX as obtained in Model Sim®



References
[1] K. Fazel and S. Kaiser, Multi-Carrier and Spread Spectrum Systems: From OFDM and MC-CDMA to LTE
and WiMAX, 2nd Edition, John Wiley & Sons, 2008, ISBN 978-0-470-99821-2
[2]. M. Ergen, Mobile Broadband - Including WiMAX and LTE, Springer, NY, 2009 ISBN 978-0-387-68189-4
[3]. Carl Weinschenk (April 16, 2010). "Speeding Up WiMax". IT Business Edge. "Today the initial WiMax
system is designed to provide 30 to 40 megabit-per-second data rates."
[4]. Wimax Forum Industry Research Report
http://www.wimaxforum.org/sites/wimaxforum.org/files/page/2011/03/Monthly_Industry_Report_March20
11.pdf
[5] Synthesis of band-limited orthogonal signals for multi-channel data transmission, Bell System Technical
Journal 46, 1775-1796.
[6] MATLAB 7.8.0 (R2009a), Help. s.l. : MathWorks, Inc., 2009.
[7] van Nee, R. and Prasad, R. OFDM for Wireless Multimedia Communications. s.l. : Artech House, 2000.
[8] Charan Langton. OFDM. Intuitive Guide to Principles of Communications. http://www.complextoreal.com/.
[9] Amalia Roca Persiva . Implementation of a WiMAX simulator in Simulink. Institute of Communications &
Radio-Frequency Engineering, Vienna University of Technology. Vienna : s.n., 2007. Master Thesis.


International Journal of Computational Engineering Research(IJCER)

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to International Journal of Computational Engineering Research(IJCER) (20)

International Journal of Computational Engineering Research(IJCER)