WiMAX: A Developer's Perspective

By: Dr. Sankarnarayan Jagannathan, Tata Elxsi Ltd
( 1 Mar 2008 )

Communication touches our daily lives in myriad ways: telephone, radio, TV, computer, and newspaper, etc. As the distance between communicators grows, it becomes economically unfeasible to connect information sources and information users using co-axial cables and fiber optic links (Figure 1).

Wireless provides feasible technology for access to remote, difficult-to-reach areas cost-effectively. WiMAX (Worldwide Interoperability for Microwave Access)—or IEEE 802.16x or Wireless MAN—provides such broadband wireless access over larger areas than 802.11 (Wireless LAN) at broadband speeds.

WIMAX ARCHITECTURE
The IEEE Working Group 16 defined two access options for a WiMAX network: fixed (IEEE 802.16TM- 2004) and portable (IEEE 802.16eTM). In the fixed option, access is provided through a fixed antenna as in a satellite television subscriber station.

In portable option, the subscriber stations are very similar to IEEE 802.11 Wi-Fi stations. The more architecture details are in [1,8].

HOW WIMAX WORKS
WiMAX requires two main components to form an operational network (Figure 2):
- Base Station (BS), which can serve as a repeater or can be connected to the Internet backbone
- Subscriber/end user, enjoying broadband wireless access through the base station

PROTOCOL STACK
The IEEE 802.16 protocol reference model has three planes: user, control and management as shown Figure 3. The IEEE standard 802.16TM-2004 deals with user and control planes. It defines two layers in these planes: Medium Access Control layer (MAC) and Physical layer (PHY). The MAC layer has three sub-layers: Service-Specific Convergence Sublayer (CS), Common Part Sub-layer (MAC CPS) and Security Sub-layer.

The CS provides the required adaptation for the up-layer incoming traffic while MAC CPS makes available key link layer functions for solving several broadband wireless communication issues.

Protocol Stacks as per IEEE 802.16 (MAC and PHY)

1. Randomizer
Randomizer combines user data with known synchronization frame data bits. The incoming data is first XOR-ed with the synchronized frame data generated from PN sequence generator. The resulting bit will be non-correlated. Same operation is performed at the receiver side.

2. Reed Solomon encoder
This functionality generally relates to polynomial-generated error correction and detection for both encoding and decoding methods.

3. Convolution encoder
A convolutional encoding process depends on past history/memory of the input symbols. The memory of the encoder is characterized by its state and represented by a v-bit binary number. For every "m" input bits, the encoder outputs "n" bits based on the "m" input, and "v" state bits, and then transitions to the next state. The code rate for convolution encoder is defined by R=m/n<1.

4. Interleaver
The performance of the forward error correction (FEC) code can be improved using interleaver. Interleaver involves permutation of a coded bit stream such that, while leaving the encoder, the adjacent bits are as widely separated as possible in various domains during transmission along the channel.

5. Quadrature Phase Shift Keying (QPSK) modulation
QPSK can encode two bits per symbol; the encoded symbols are mapped on constellation using gray code. The phase of the carrier wave is modulated to encode bits of digital information in each phase change.

6. Orthogonal Frequency Division Multiple Access (OFDMA)
The transmission is consists of S/P converts to rearranging the symbol to perform the IFFT operation. The parallel converted symbols are arranged to subscriber stations, after assigning the symbols to the sub channel, IFFT operation is performed. After this, the output of the IFFT is converted from serial to parallel operation. The final resulting data is transmitted.

7. Ranging
The BS provides the MS with an initial notification of a periodic ranging time that occurs during a sleep time interval. The MS is to perform the ranging process during this time. The initial notification is included in the first message, indicating whether the MS should terminate sleep mode to receive DL data. The second message is transmitted to the MS as part of the ranging process such that MS performs a plurality of ranging processes within the sleep time interval

8. MAC management
A method for managing the MAC message is provided, which supports reservation of dynamic resources for upstream data traffic in broadband cable systems. Three specialized MAC management messages—Dynamic Session Addition, Dynamic Session deletion, and Dynamic Session Acknowledgement—control the setting of a filter specification in a cable modem. The RSVP (Resource ReSerVation Protocol) protocol is adapted for use at the OSI protocol layer of a typical cable modem, leading to efficient network bandwidth/resource allocation.

9. Priority queue
The queues includes a low priority queue for Best Effort (BE) traffic, a medium priority queue for streaming data such as video pictures, and a high priority queue for voice traffic—serviced in that order of priority. The low and medium priority traffic is handled on a weighted round-robin basis, and high priority traffic takes precedence over both.

10. Scheduler
A packet scheduler provides a high degree of fairness in scheduling packets associated with different sessions to minimize packet delay during packet transmission from a plurality of sessions with, possibly, different requirements and transfer rates. On receipt of a packet, scheduler assigns a start time to it based on whether the session has pending packets or not, the values of the end time of the previous packet in the session, and the packets' arrival time. The scheduler determines the end time of the packet based on the transfer time required for the packet based on packet length/rate, and by adding the Transfer time to the packet start time.

11. Fragmentation
A Wireless Transmit/Receive Unit (WTRU) includes a data de-fragmentation unit that de-fragments any fragmented data received by the WTRU. The WTRU does not transmit fragmented data. The WTRU includes a processor, a data fragmentation unit, a transmitter and a fragmentation selection unit. The fragmentation and re-assembly of tunneled packets are handled in the hardware pipeline to achieve line-rate processing of the traffic flow without any need for additional store-and-forward operations typically provided by a host processor/co-processor. The hardware pipeline may perform packet fragmentation and reassembly, performing segment-by-segment crypto using encrypted tunnels. A network device implementing packet fragmentation can include a hardware pipeline that fragments packets between the input packet memory of the device and the MAC and encrypts/tunnels outgoing fragments.

12. Concatenation
This comprises of a wireless network, including a transmitting device and multiple receiving devices. The transmitting device is configured to receive multiple data packets, prepare a preamble, prepare a signal field for each data packet, and broadcast as a concatenated packet the preamble, the first signal field, the first data packet, the second signal field, and the second data packet.

SIMULATION AND MODELING
Modeling and simulating WiMAX represents a big challenge. To contend with this, we need to intersect two simulation approaches typically applied independent of each other: signal simulation is used for the air interface performance of the physical layer and protocol simulation is used for up-layer protocol performance.

Therefore, a tool to evaluate WiMAX performance should, preferably, integrate both environments in the same software. This allows for a more complete interaction among models as different from mere parameter exchange. Another advantage is that detailed models can be replaced by simple parameters if a more simple analysis is required. Finally, due to IEEE 802.16 complexity, modeling approaches should take care to offset the drawbacks relating to model complexity, model details level, model scope and computational performance.

The WiMAX subscriber station physical layer is modeled and simulated as per the HiperMAN standards. This model has to be coded and then downloaded on to the FPGA or DSP chip for testing. In like manner, the MAC layer is developed such that it can be downloaded on to the same chip. This way an entire subscriber station is built. The advantage of using Simulink for model-based design is that redundancies and minor errors that could possibly occur at different stages of modeling can be detected and corrected. Also, different variations of the given standard can be tested even before these are tested on the chip, reducing hardware and labor costs.

This model-based design of the WiMAX physical layer can serve as the prime model for chip design since it contains all the required parameters. Each block is coded in C language before being tested and recoded into any hardware description language (HDL).

LOOKING AHEAD
There is soaring demand for wireless broadband access, and an ever-widening range of applications today require fixed, nomadic, portable and mobile data access as well as fixed and mobile voice services, and content streaming. WiMAX is committed to meeting the requirements of all these applications.

REFERENCES
1. WiMAX Forum: Technical specifications white paper, "Broadband Radio Access Networks (BRAN) HiperMan Physical (PHY) Layer"
2. WiMAX Forum: Technical specifications white paper, "OFDM Processing Survey"
3. The Mathworks Inc.: Communication Blockset for use with Simulink
4. IEEE Communication Magazine, Vol. 43, No.1, 2, January 2005
5. IEEE Networks Magazine, Vol. 27, No.2, January 2005
6. IEEE Wireless Communication Magazine, Vol. 43, No. 1, January 2005
7. Scientific American Journal, Vol. 25, No. 6, March 2004
8. www.wimaxforum.com
9. www.ieeeexplore.com
10. www.acm.com
11. www.mathworks.com
12. www.edaboard.com
13. www.altera.com
14. www.drdobbs.com


About the Author
Dr. Sankarnarayan Jagannathan is Head of Patents and Publications, Tata Elxsi Ltd, Bangalore, India.

Click here for Illustrations:

Figure 1, Figure 2, Figure 3