5. Simulating Channel Allocation Performance using OPNET

5.1      Introduction

In a cellular system, the amount of spectrum available for a wireless operator is usually limited.  Consequently, the channels available are limited and need to be reused.  Channel reuse causes interference and which degrades the Signal to Noise Ratio (SNR) of received packets.  Poor SNR will lower the data throughput in a data network or increases the blocking and dropping probabilities in a voice network.  An effective way to reduce this interference is to have good channel allocation.

Adjacent channel interference and co-channel interference are a result of channel reuse.  Adjacent channel interference occurs if two adjacent frequency bands transmit close together and this can be avoided if sufficient guard band is available between these two adjacent frequency bands.  Co-channel interference occurs when two base stations using the same frequency band transmit at the same time.  Co-channel interference can be reduced if the two base stations are a sufficient distance apart.  To minimise these interferences, the wireless channels need to be strategically allocated to each base station. 

The task of channel allocation is to assign N channels to M base-stations given a constraint in the form of a compatibility matrix, C.  The compatibility matrix is a M ´ M matrix stating the channel separation between any two base-stations in the network to ensure that two base-stations using the same channel are a sufficient distance apart.  The channel allocation problem can be transformed to a graph-colouring problem [1] and thus it is a NP-complete problem [2] .  To allocate a channel, information on the channel usage or interference environment is required.  The performance of a channel allocation scheme is proportional to the amount of information it acquires and the effectiveness with which this information is used [13] .  The Channel Allocation methods are described in a Channel Allocation Matrix, which classifies the methods according to the technique employed to acquire network information.  The Channel Allocation Matrix is shown in Figure 5.1.

Figure 5.1: Channel Allocation Matrix

The vertical axis in Figure 5.1 is a measure of the centralisation required in the channel allocation method.  The degree of centralisation is defined as being proportional to the number of base-stations required to communicate with a central controller in order to allocate a channel.  A fully centralised system requires every base-station in the entire network to communicate with a central controller while in a fully distributed system, the base-station can make the decision to allocate a channel on its own.  The more centralised the system the greater is the amount of signaling required, which causes high packet or call set-up delay and may result in system instability.  The more distributed the system is, the less global knowledge is present at each base-station and so the decision is based on partial knowledge and usually the allocation is made to benefit itself.  The horizontal axis in Figure 5.1 represents the quantity of measurements (e.g. interference power or SNR) made by the base-station and/or subscriber’s unit prior to a channel allocation.  Measurement adds to the complexity of the process and needs to be performed quickly to minimise packet delay.  In a non-measuring scheme, a-priori knowledge of the network such as the reuse distance and the compatibility matrix are used.  The Channel Allocation Matrix can be divided into four quadrants: Distributed Non-measurement, Centralised Non-measurement, Distributed Measurement and Centralised Measurement.  Citations to channel allocation methods in each quadrant are shown in Figure 5.1.

6.2      System Description

The project investigates channel allocation methods for Broadband Fixed Wireless Access (BFWA) system.  The BFWA system under consideration in this project consists of the Control Server (CSVR), the Access Point (AP) and the Subscriber Unit (SU) and they are arranged as shown in Figure 5.2.  The SU is mounted on the subscriber’s site and uses a directional antenna with a horizontal beamwidth of 20° to communicate with its corresponding AP.  Each AP has an antenna with a horizontal beamwidth of 60°.  Several APs can be connected to a CSVR where configuration and authentication are provided.  The CSVRs are further connected using high-speed networks to a central control facility for billing, operations and maintenance monitoring.

Figure 5.2. Broadband fixed wireless access network components.

The SUs uses a Packet Reservation Multiple Access scheme to gain access to an AP.  Two way communication is achieved using Time Division Duplex.  A single MAC frame structure is illustrated in Figure 5.3.

Figure 5.3. A single MAC frame structure.

The AP broadcasts data packets to its SUs using the Downlink portion of the MAC frame and the SUs use the Uplink portion of the MAC frame to transmit data packets to their AP.  The packet throughput and packet delay of this MAC scheme is simulated using OPNET.  The performance is compared with two MAC schemes, namely the ALOHA and Slotted-ALOHA.  The results of the simulation are published in [8].

6.3      Simulation Models

The AP and SU of the BFWA system described in Section 1.2 are modelled using OPNET.  Channel Allocation methods lying in the lower two quadrants of the Channel Allocation Matrix are investigated for a BFWA system.  Both the AP and SU are capable of changing the transmit frequency during a simulation and this characteristic is ideal for a DCA simulation.  The receiver’s frequency can also be changed during a simulation and this is used for interference power measurements.  In OPNET the wireless channel is modelled using a 14 pipeline stage.  Some of these stages need to be modified for the simulation.  For example, the propagation model was changed from the default free space propagation model to a Random Height propagation model that is suitable for a fixed wireless network [14] .  The antenna pattern in OPNET can be designed according to the users specification.  The directional and sectored antennas used by the SU and AP respectively are thus modelled using OPNET’s Antenna Pattern editor.

The simulation is performed using self-similar traffic that is typical of a packet data network.  An ON-OFF packet generator node was created and the self-similar traffic is modelled using a Pareto distribution [ 15], which is available in OPNET distribution package.

The packets in OPNET are capable of storing information such as the Signal to Noise Ratio and the received power, which are calculated by the simulation kernel and radio pipeline stages.  The results collected are exported into an Excel spreadsheet for analysis.

Broadband wireless access uses high bit rates for communication.  For example, in a 25 Mbps system, a burst of 1000 bits will take 40 ms.  In this model, each packet arrival will generate a series of events such as requesting a timeslot and going through the 14 radio pipeline stages.  If the data rate is set to 25 Mbps, the model will require a significant amount of time just to run one minute of simulation.  The concept of the packet frame unit is introduced in which the data rate is normalised to the average packet length (e.g. in bits or one cell in an ATM system).  Thus one packet will occupy one second or a packet frame.  Hence one second of simulation time in OPNET is equivalent to one packet frame of simulation time.  By doing this, the effective data rate can be obtained from the channel utilisation (i.e. the number of bits per second) statistics available in OPNET.

Several DCA schemes such as Random Channel Assignment (RND), Channel Segregation [5] and Least Interfered [6] are modelled for the BFWA system and simulated under typical traffic.  New DCA schemes were developed and simulated to compare with the existing DCA schemes.  The new schemes fall into the lower quadrant of the Channel Allocation Matrix and they are DCA using Game Theory GA ([9] , [10] ), DCA using Genetic Algorithm GA ([11] , [12] ) and FCA using Genetic Algorithm FCA-GA [13] .  The results are presented in the cited papers. 

The disadvantage of using OPNET is that the simulation requires a lot of processing power and can be very time consuming particularly for network with a large number of transmitter and receivers.  This is mainly due to the detailed radio pipeline stage that OPNET uses since every packet transmitted is required to go through these stages.  The simulation time can be reduced by using parallel processors.

6.4      Conclusion

A wireless point to multi-point simulation has been developed using OPNET to study the performance of various channel allocation schemes for a PRMA based broadband fixed wireless access system.  New channel allocation methods ([9] - [13] ) have been developed and analysed using OPNET.  OPNET has shown itself suitable for this study because it is capable of simulating the radio propagation channel and is flexible enough for the user to accurately model the system.

6.5 References

[1]           J. Arthur Zoellner and C. Lyle Beall, “A Breakthrough in Spectrum Conserving Frequency Assignment Technology,” IEEE Transactions on Electromagnetic Compatibility, vol. EMC-19, no. 3, pp. 313-319, August 1977.

[2]           Michael R. Garey and David S. Johnson, Computers and Intractability: A Guide to the Theory of NP-Completeness.  San Francisco: W. H. Freeman and Co., 1979.

[3]           Sanjiv Nanda and David J. Goodman, “Dynamic Resource Acquisition: Distributed Carrier Allocation for TDMA Cellular Systems,” in Proc. IEEE Global Telecommunications Conference (GLOBECOM’91), December 2-5, 1991, Phoenix, Arizona, vol. 2, pp. 883-889.

[4]           Kumar N. Sivarajan and Robert J. McEliece, “Dynamic Channel Assignment in Cellular Radio,” in Proc. IEEE 40^th Vehicular Technology Conference, May 6-9, 1990, Orlando, Florida, pp. 631-637.

[5]           Yoshihiko Akaiwa and Hidehiro Andoh, “Channel Segregation – A Self-Organized Dynamic Channel Allocation Method: Application to TDMA/FDMA Microcellular System,” IEEE Journal on Selected Areas in Communications, vol. 11, no. 6, pp. 949-954, August 1993.

[6]           Matthew Cheng and Li Fung Chang, “Wireless Dynamic Channel Assignment Performance Under Packet Data Traffic,” IEEE Journal on Selected Areas in Communications, vol. 17, no. 7, pp. 1257-1269, July 1999.

[7]           Osman Koyuncu, Sajal K. Das and Hakan Ernam, “Dynamic Resource Assignment Using Network Flows in Wireless Data Networks,” in Proc. IEEE Vehicular Technology Conference 1999, May 16-19, 1999 Houston Texas, pp. 1-5.

[8]           Shin Horng Wong and Ian J. Wassell, “Performance Evaluation of a Packet Reservation Multiple Access (PRMA) Scheme for Broadband Fixed Wireless Access,” in Proc. London Communications Symposium 2001, Sep. 10-11, 2001, London, pp. 179-182.

[9]           Shin Horng Wong and Ian J. Wassell, “Application of Game Theory for Distributed Dynamic Channel Allocation,” in Proc. IEEE 55^th Vehicular Technology Conference, Spring 2002, May 6-9, 2002, Birmingham, AL, USA, vol. 1, pp. 404-408.

[10]      Shin Horng Wong and Ian J. Wassell, “Distributed Dynamic Channel Allocation Using Game Theory for BroadbandFixed Wireless Access," in Proc. 2002 International Conference on Third Generation Wireless and Beyond, May 28-31, 2002, San Francisco, pp. 304-309

[11]      Shin Horng Wong and Ian J. Wassell, "Dynamic Channel Allocation for Interference Avoidance in a Broadband Fixed Wireless Access Network," in Proc. 3^rd International Symposium on Communication Systems Networks and Digital Signal Processing (CSNDSP02), July 15-17, 2002, Staffordshire University, UK, pp. 352-355.

[12]      Shin Horng Wong and Ian J. Wassell, "Dynamic Channel Allocation Using a Genetic Algorithm for a TDD Broadband Fixed Wireless Access Network," in Proc. IASTED International Conference in Wireless and Optical Communications, July 17-19, 2002, Banff, Alberta, Canada, pp. 521-526

[13]      Shin Horng Wong and Ian J. Wassell, “Channel Allocation for Broadband Fixed Wireless Access,” in Proc. 5^th International Symposium on Wireless Personal Multimedia Communications (WPMC02), Oct 27-30, 2002. Honolulu, Hawaii.

[14]      D. Crosby, Propagation Modelling for Directional Fixed Wireless Access System. Ph.D. dissertation, University of Cambridge, 17 November 1999.

[15]      Walter Willinger, Murad S. Taqqu, Robert Sherman and Daniel V. Wilson, “Self-Similarity Through High-Variability: Statistical Analysis of Ethernet LAN Traffic at the Source Level,” IEEE/ACM Transactions on Networking, vol. 5, no. 1, pp. 71-86, February 1997.