• 沒有找到結果。

Interference Avoidance Transmission Scheme for SOP

4 Interference Avoidance Transmission Scheme for IEEE 802.15.3a

4.3 Simultaneously Operating Piconets (SOP)

4.3.2 Interference Avoidance Transmission Scheme for SOP

According to Specification of IEEE 802.15.3, the PNC should periodically listen in the current channel to detect interference resulting from the presence of other 802.15.3a piconets or the presence of other wireless networks. When there are other piconets detected by the PNC, four methods that are available to mitigate the interference between the two piconets are listed as follow

1. The PNC may join the other piconet to form a child piconet.

2. The PNC may join the other piconet to form a neighbor network.

3. The PNC may change channels to one that is unoccupied.

4. The PNC may reduce the maximum transmit power in the piconet.

The first two methods can complete avoid the interference due to two piconets become one piconet. However, the last two methods are not suitable for the MB OFDM system. Because signals of interference piconet exist in all channels and they still influence the desired piconet even though the desired piconet reduces the transmit power. To further improve performance in presence of other piconet, we propose an interference avoidance transmission scheme. The MB OFDM system employs the preamble to aid receiver algorithms related to synchronization, carrier-offset recovery, and channel estimation. We can exploit the property to find other piconet, when other piconet has existed in the frequency slots of the desired piconet based on the specific time-frequency interleaving pattern.

First, the desired PNC will listen to channels according to the specific time-frequency interleaving pattern. The desired PNC will calculate the cross-correlation of two successive signals in the same frequency slot. The lth received signal of an L-length window integrate-and-dump cross-correlator is

1 H

l l l

P = r r+ (4.3)

where rl =[ , ,...,r r0 1 rL1]T and [ ]⋅H denotes Hermitian transpose. Further, normalizing P produces the correlation coefficient l

1 l

l l l

P ρ R R

+

= (4.4)

where Rl = r r and the range of lH l ρ is constrained to [0 1]. The receiver will l declare the interference detection at the frequency slot when some threshold of correlation Tc is exceeded, that is, whenρl >Tc. The commonly used value of Tc = 0.8 [37].

When other piconet has existed in the frequency slots of the desired piconet, the

desired PNC has to know how many portions of each frequency slot are overlapped with the frequency slot of the interference piconet. Consequently, we divide the each desired frequency slot into several smaller time-frequency slots, as shown in Figure 4.5. Then, we also calculate the correlation coefficient of each time-frequency slot in the desired frequency slots which have been detected the presence of interference piconet. Similarly, if the time-frequency slot in the desired frequency slots is overlapped with the frequency slost of the interference piconet, the threshold of correlation Tc is exceeded. By counting the number of time-frequency slots covered with frequency slots of the interference piconet, the PNC can find out the portions of each frequency slot overlapped with the frequency slots of the interference piconet.

According to the above result, the PNC can decide the shift time which makes the each frequency slot of two piconets totally separate or cover with each other. After adjusting the time of the desired frequency slot in the time axis, there are only two desired frequency slots with other piconet interference. The other frequency slots are interference-free for the desired piconet. Additionally, if the interference piconet has existed in this area, the desired piconet cannot occupy the frequency slots of the interference piconet. For this reason, the desired piconet will close down the two frequency slots jammed by other piconet. The devices in the desired piconet will transmit signals with the remainder of the frequency slots. Figure 4.6 presents the flow chart of the interference avoidance for desired piconet

For example, the desired piconet uses the time-frequency interleaving pattern {f1, f3, f2, f1, f3, f2, repeats}, whereas the interference piconet uses the time-frequency interleaving pattern {f1, f2, f3, f1, f2, f3, repeats}. As shown in Figure 4.4, the signal of the interference piconet and the desired piconet are not time-aligned due to the lack of coordination of transmissions among two piconets.

Collisions will occur between two piconets. At worst case, there are four frequency

slots occurred collisions between two piconets, so that the desired piconet may not decode the signal correctly. By means of the correlation coefficient of each time-frequency slot, the desired piconet shifts its own specific time-frequency interleaving pattern in the time axis. Therefore, there are only two frequency slots overlapped with the interference piconet, as shown in Figure 4.7. And the desired piconet turns off the two frequency slots. Because the MB OFDM system improves the performance of SOP by all three techniques: spreading, convolutional code and TFIC. We consider the binary information bits of desired piconet are encoded with a coding rate of 1/3 in 106.7 Mbps mode as shown in Figure 4.8. Each 200 coded-bit output of each generator polynomial of the encoder is interleaved, mapped into one of QPSK constellation points and OFDM-modulated by 128-points IFFT. Each OFDM symbol is transmitted twice at different frequency slots based in the specific time-frequency interleaving pattern. When 1st and 4th frequency slots have been turned off, the transmission method which keeps the coding rate (1/3) and loses partial frequency diversity is shown in Figure 4.9. The first 200 coded-bit can be received from 2nd OFDM symbol and the second 200 coded-bit can be received from 3rd OFDM symbol. In addition, the third 200 coded-bit can be received from coherently combining 5th and 6th OFDM symbols. Consequently, the third 200 coded-bit can maintain the frequency diversity. Then, the receiver can decode the desired signal correctly.

4.4 Computer Simulations

Computer simulations are conducted to evaluate the performance of the interference avoidance transmission scheme in the MB OFDM system. Because the MB OFDM system uses spreading, coding and TFICs to improve the performance of SOP, we consider the 106.7 Mbps data rate mode which has enough redundancy to allow the receiver to decode the desired signals correctly in the presence of other piconets.

In the first simulation, the performance of the interference avoidance transmission scheme is investigated with different numbers of time-frequency slots in each frequency slot. The ideal case means that the transmitter conveys signals with four interference-free frequency slots and the signals are not jammed by other piconets In addition, the MRC case means that the transmitter conveys signals with six frequency slots and the receiver decodes the signals by MRC. The BER performances of the proposed method with different number of time-frequency slots in each frequency slot in the CM1-2 channels are shown in Figures 4.10-4.11. They show that the BER associated with eight time-frequency slots in each frequency slot is close to the ideal case. This means that the proposed method with eight time-frequency slots in each frequency slot can avoid most of the interference from other piconets. In addition, the BER associated with two or four time-frequency slots in each frequency slot have an error floor, implying the two and four divisions are not enough. This is because the more time-frequency slots are divided in a frequency slot, the more accurately we can shift to the desired position, and the less severely the collision due to the inaccurate shift will occur.

In the second simulation, the performance of the interference avoidance transmission scheme is investigated with different SIR. Each frequency slot is

divided into eight time-frequency slots. The BER performances of the proposed method with different SIR in the CM1 channel are shown in Figure 4.12. The curve illustrates that the proposed method with SIR = −10 dB only loses 2.5 dB for a BER of 104 compared with the ideal case. This means that the proposed method can effectively mitigate the interference even if the interference is larger than the desired signal.

4.5 Summary

In the MB OFDM system, TDMA is used for devices within the same piconet.

However, the conventional multiple accesses are not suitable for multi-piconet environments due to the lack of coordination of transmissions among different piconets. In order to improve the performance under SOP, the MB OFDM system uses three methods, spreading, convolutional code, and TFICs, to achieve bandwidth expansion to alleviate the interference. In practice, the performance of SOP will be seriously degraded when the power of other piconets is high. To further improve the performance of SOP, we proposed the interference avoidance transmission scheme.

When other piconets are present in the same area, the desired piconet will convey the signals in the interference-free frequency slots. By partially abandoning frequency diversity, the transmitter can keep the original coding rate which is shown to be able to significantly improve the performance of SOP. The simulation results indicate that the proposed scheme effectively alleviates the interference from other piconets and the resulting BER performance is close to the ideal case.

PNC/DEV DEV DEV

DEV

DEV

beacon data

Figure 4.1: The 802.15.3 piconet elements

Superframe m-1

Superframe m Superframe m+1

Beacon m

Contention Access

Period

Contention Free Period MCTA

1

MCTA

2 CTA 1 CTA 2 CTA n-1 CTA n

Figure 4.2: The 802.15.3 piconet superframe

Coding

Figure 4.3: Pictorial representation of bandwidth expansion for the MB OFDM system

Figure 4.4: Collision property of two time–frequency interleaving codes for two piconets Interference signal desired signal

Frequency slot

Figure 4.5: Illustration of time-frequency slots in each frequency slot

Figure 4.6: Flow chart of the interference avoidance transmission scheme

Frequency Slot

Time-Frequency Slot

Detect interference

Calculate number of overlapped

time-frequency slots

Shift specific TFI pattern

Turn off two jammed frequency slots

Yes

Convey signals according to

TFI pattern Detect No interference

Calculate number of overlapped

time-frequency slots

Shift specific TFI pattern

Turn off two jammed frequency slots

Yes

Convey signals according to

TFI pattern No

Figure 4.7: The ideal collision situation for no coordination of transmissions among two piconets

Figure 4.8: Example for the 106.7 Mbps mode of the MB OFDM system

Band 3

Interference signal desired signal

Frequency slot Interference signal desired signal

Frequency slot

Figure 4.9: Example of transmission scheme for the 106.7 Mbps data rate mode of the MB OFDM system

Figure 4.10: Coded BER versus Eb/N0 for the 106.7 Mbps data rate mode of the MB OFDM system in the CM 1 channel with different number time-frequency slots (SIR = 0 dB)

Combining (frequency diversity) Original coding rate is 1/3

G1 G2

G3

Coding rate 1/3, loss partial frequency diversity Combining

(frequency diversity) Original coding rate is 1/3

G1 G2

G3

Coding rate 1/3, loss partial frequency diversity

-10 -5 0 5 10 15 20

Figure 4.11: Coded BER versus Eb/N0 for the 106.7 Mbps data rate mode of the MB OFDM system in the CM 2 channel with different number time- frequency slots (SIR=0 dB)

Figure 4.12: Coded BER versus Eb/N0 for the 106.7 Mbps data rate mode of the MB OFDM system in the CM 1 channel with different SIR

-6 -4 -2 0 2 4 6 8 10 12

10-4 10-3 10-2 10-1 100

ideal 0 dB -5 dB -10 dB

Bit Error Rate

Eb/N0 (dB)

-10 -5 0 5 10 15 20

10-4 10-3 10-2 10-1 100

2 time-frequency slot 4 time-frequency slots 8 time-frequncy slots ideal

MRC

Bi t Err or R at e

E

b

/N

0

(dB)

Chapter 5 Conclusion

In this thesis, we propose a MB OFDM system incorporating decision-aided ICI canceller (DAIC) and interference avoidance transmission scheme. The DAIC solves the impact of ICI induced by long delay spread channels and has a lower complexity than the MMSE equalizer. The interference avoidance transmission scheme effectively mitigates degradation of performance when other piconet is present in the same area. In Chapter 2, the transmitter architecture of IEEE 802.15.3a MB OFDM system is introduced. Although the architecture is similar to that of conventional OFDM systems, the ZPP OFDM, spreading, and frequency hopping are three major characteristics different from conventional OFDM systems. In Chapter 3, we introduced the S-V model which is suitable for UWB channel environment. Then, based on the MB OFDM technical specification, we construct the receiver architecture. Synchronization and channel estimation algorithms for the MB OFDM receiver are established. The ZPP OFDM transceiver effectively eliminating the ripples in the PSD of transmitted signal is also discussed. Besides, the influence of ICI resulting from the long delay spread channel is described. We also discuss three feasible methods to improve the effect of ICI. For moderate and low data rate modes of the MB OFDM system, the receiver can mitigate the impact of ICI by MRC method. Besides, the other methods are workable for high data rate modes without

any spreading. However, the MMSE equalizer is too complex to be adopted for the MB OFDM system. The DAIC method can significantly reduce the error floors induced by the ICI and have lower complexity than MMSE equalizer. In addition, the LS channel estimation is not suitable for the CM4 channel. Because the two OFDM training symbols suffer the same influence by ICI, the LS channel estimation cannot perform reliably under the CM4 channel. We develop the DAIC method to estimate the channel impulse response. The simulation result indicates that the DAIC channel estimation effectively mitigates the effect of ICI and is more accurate than the LS channel estimation.

In Chapter 4, we first review the multiple access techniques including FDMA, TDMA, and CDMA. Then, the multiple access for intra- and inter-piconet are presented. A number of devices in a piconet share the fixed communication resource by TDMA. In the MB OFDM system, the four unique time-frequency interleaving codes are adopted to specify different piconets. However, the collisions still occur when two piconets exist at the same area. The proposed interference avoidance avoids the interference by abandoning the collided frequency slots. Then, transmitter will convey the signal in remainder frequency slots. The simulations indicate the proposed method can effectively avoid the interference and the performance of BER is close to the interference-free performance of BER.

The study presented in the thesis has discussed the receiver design for MB OFDM system and its efficacy has been verified using simulation platform. Although we present workable methods for advanced MB OFDM system, there are some challenges in implementing MB OFDM system, such as hardware and RF.

Furthermore, we can investigate the performance of the interference avoidance transmission scheme in the network simulator, like NS2 and OPNET. By cross layer design, we can improve the performance of SOP further.

In practice, there are two competing UWB specifications: MB OFDM and DS-UWB systems. There are different superiorities of some system over the other. In the MB OFDM system, the obvious advantage is its capability to capture multipath energy with a simple FFT. In contrast to CDMA, the Rake correlator fingers should be used to exploit multipath diversity. The Rake correlators in CDMA based UWB receivers lead to a more complex receiver with higher power consumption. In addition, OFDM enables the designer to adapt the system to avoid using some specific bands to comply with other regulations set forth by other countries. However, the MB OFDM system still has some problems of conventional OFDM systems, such as the complexity of the transmitter and high peak-to-average power ratio (PAPR).

Besides these conventional problems, the MB OFDM system has to overcome the considerably large power consumption of power amplifier and multi-nanosecond hopping time between the subbands. In spite of the advantages and drawbacks, there are some design challenges for both systems, such as frond-end LNA/mixer which have satisfied the requirements over the 7.5 GHz bandwidth set forth by FCC for UWB communications. The other notable challenge is the architectures of ADC in the UWB transceiver. They are too expensive and power consuming for UWB systems.

Although both groups are focusing on having the IEEE 802.15 standards group adopt and manage their UWB specification as a standard, it's still unclear at this point which of the specifications will win. It's very possible that both will eventually become part of the same standard, similar to 802.11a and 802.11b/g is with WLANs.

Bibliography

[1] L.Q. Yang and G.B. Giannakis, “Ultra-wideband communications: an idea whose time has come,” IEEE Signal Processing Magazine, Vol. 21, pp. 26-54, Nov. 2004

[2] G.R. Aiello and G.D. Rogerson, “Ultra-wideband wireless systems,” IEEE Microwave Magazine, Vol. 4, pp.36-47, June 2003

[3] FCC, “First report and order, revision of part 15 of the commission’s rules reguarding ultra-wideband transmission systems,” ET Docket, pp. 98-153, Feb.

2002

[4] IEEE 802.15WPAN High Rate Alternative PHY Task Group 3a (TG3a) [Online]. Available: http://www.ieee802.org/15/pub/TG3a.html

[5] WLAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications, ANSI/IEEE 802.11, 1999

[6] Local and Metropolitan Area Networks, IEEE 802.3, 2002 [7] Universal Serial Bus Specification, Revision 2.0, 2000

[8] Standard for a High-Performance Serial Bus, IEEE 1394, 1995

[9] R. Fisher et al., “DS-UWB Physical Layer Submission to 802.15 Task Group 3a,” IEEE 802.15-04/0137r3, Motorola, Inc. et al., July 2004

[10] R. Fisher et al., “DS-UWB Proposal Update for IEEE P802.15 Working Group for Wireless Personal Networks (WPANs),” IEEE 802.15-64/04140r7, Motorola, Inc. et al., July 2004

[11] R. Fisher et al., “Merger#2 Proposal Update for IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs),” IEEE 802.15-64/022r0, Motorola, Inc. et al., Jan. 2004

[12] A. Batra et al., “Multiband OFDM physical layer proposal for IEEE 802.15 Task Group 3a,”Multi-bandOFDMAlliance, Sep.2004

[13] A. Batra et al., “Design of a multiband OFDM system for realistic UWB Channel Environments,” IEEE Transactions on Microwave Theory and Techniques, vol. 52, pp. 2123-2138, Sep. 2004

[14] V.S. Somayazulu, J.R. Foerster, and S. Roy, “Design challenges for very high data rate UWB systems,” Systems and Computation Conference in Processing.

Asilomar Signal, pp. 717-721, Nov. 2002

[15] A.F. Molisch, J.R. Foerster, and M. Pendergrass, “Channel models for ultrawideband personal area networks,” IEEE Wireless Communications, Vol.

10,pp. 14-21, Dec. 2003

[16] D. Cassioli, M.Z. Win, and A.F. Molisch, “The ultra-wide bandwidth indoor channel: from statistical model to simulations,” IEEE Journal on Selected Areas in Communications, Vol. 20, pp. 1247-1257, Aug. 2002

[17] IEEE 802.11-97/96, Naftali Chayat, Sep. 1997.

[18] A. Saleh and R. Valenzuela, “A Statistical Model for Indoor Multipath Propagation,” IEEE Journal on Selected Areas in Communications, vol. 5, no. 2, pp. 128–137, Feb. 1987.

[19] H. Hashemi, “Impulse Response Modeling of Indoor Radio Propagation Channels,” IEEE Journal on Selected Areas in Communications, vol. 11, no. 7, pp. 967–978, Sept. 1993.

[20] J. Terry and J. Heishkala, OFDM wireless LANs: A theoretical and practical guide, Indiana: SAMS, 2001

[21] B. Muquet et al., “Cyclic prefixing or zero padding for wireless multicarrier transmissions,” IEEE Transactions on Communications, Vol. 50, pp. 2136-2148, Dec. 2002

[22] A. Scaglione, G.B. Giannakis, and S. Barbarossa, “Redundant filterbank precoders and equalizers. I. Unification and optimal designs,” IEEE Transactions on Signal Processing, Vol. 47, pp. 1988-2006, July 1999

[23] A. Scaglione, G.B. Giannakis, and S. Barbarossa, “Redundant filterbank precoders and equalizers. II. Blind channel estimation, synchronization, and direct equalization,” IEEE Transactions on Signal Processing, Vol. 47, pp. 2007 - 2022, July 1999

[24] H.J. Yu, M.S. Kim; T.H. Jeori, and S.K. Lee, “Equalization scheme for OFDM systems in long delay spread channels,” 15th IEEE International Symposium on

PIMRC, Vol. 2, pp. 1297-1301, Sept. 2004

[25] S.P. Chen and T.R. Yao, “FEQ for OFDM systems with insufficient CP,” 14th IEEE Proceedings on Personal, Indoor and Mobile Radio Communications, Vol. 1, pp. 550-553, Sept. 2003

[26] C.G. Wang and Z. Zhou; “A new detection algorithm for OFDM system without cyclic prefix,” Proceedings of the IEEE 6th Circuits and Systems Symposium on Emerging Technologies, Vol. 2, pp. 453-456, June 2004

[27] S. Yi and T. Lang, “Channel equalization using one-tap DFE for wireless OFDM systems with ICI and ISI,” IEEE Workshop on Signal Processing Advances in Wireless Communications, pp. 146-149, May 1999

[28] S. Yi, “Bandwidth-efficient wireless OFDM,” IEEE Journal on Selected Areas in Communications, Vol. 19, pp. 2267-2278, Nov. 2001

[29] N. Suzuki, H. Uehara, and M. Yokoyama, “A new OFDM demodulation method to reduce influence of ISI due to longer delay than guard interval,” The 8th International Conference on Communication Systems, Vol. 1, pp. 239-244,

[29] N. Suzuki, H. Uehara, and M. Yokoyama, “A new OFDM demodulation method to reduce influence of ISI due to longer delay than guard interval,” The 8th International Conference on Communication Systems, Vol. 1, pp. 239-244,

相關文件