Transmit Power Control Suitable for Interference-Aware Channel Segregation Based Dynamic Channel Assignment

Transcription

1 Transit ower Control Suitable for Interference-Aware Channel Segregation Based Dynaic Channel Assignent Yuki Matsuura 1 Katsuhiro Tea 1 Koichi Ishihara 2 B. A. Hirantha Sithira Abeysekera 2 Tooaki Kuagai 2 and Fuiyuki Adachi 3 1, 3 Dept. of Counications Engineering, Graduate School of Engineering, Tohoku University 6-6-5, Aza-Aoba, Araaki, Aoba-ku, Sendai, , JAAN 2 NTT Network Innovation aboratories, NTT Corporation 1-1, Hikarinooka, Yokosuka-shi, , Japan E-ail: 1 {atsuura, tena}@obile.ecei.tohoku.ac.jp 3 adachi@ecei.tohoku.ac.jp 2 {ishihara.koichi, hirantha.abeysekera, kuagai.tooaki}@lab.ntt.co.jp Abstract Since the nuber of available channels is liited, the sae channel needs to be reused. Therefore, the co-channel interference (CCI) liits the transission quality. The channels should be reused so that the CCIs received at all access points (As) or base stations (BSs) are iniized. Our recently proposed interference-aware channel segregation based dynaic channel assignent (IACS-DCA) can for a stable channel reuse pattern which itigates the CCI in a distributed anner. An additional use of transit power control (TC) can further reduce the CCI. In this paper, we propose a signalto-interference power ratio (SIR) based TC schee suitable for IACS-DCA. We show, by coputer siulation, that the IACS-DCA with SIR based TC fors a stable channel reuse pattern and further iproves the outage probability of signal-to-interference-plus-noise power ratio (SINR). Keywords transit power control; channel segregation; dynaic channel assignent; co-channel interference I. INTRODUCTION The nuber of channels available for wireless networks is liited, and therefore, the sae channel needs to be reused by spatially separated access points (As) or base stations (BSs). It is iportant to reuse the sae channel so as to iniize the co-channel interferences (CCIs) received at all As or BSs. However, in wireless local area network (WAN), ost of A easures the CCI levels on the available channels and selects the best channel having the lowest CCI only when powered on (in this paper, this is referred to as the quasi-dynaic channel assignent (DCA)). However, the CCI changes in tie after A is powered on, and therefore the channel selected by quasi- DCA cannot be the best all the tie. In order to adapt to the changing CCI environent, DCA [1]-[3] needs to be eployed. An application of DCA to wireless networks has been studied extensively in the literature [4]-[6]. One proising DCA is channel segregation based DCA (CS-DCA) [7],[8]. Recently, we proposed an interference-aware CS-DCA (IACS-DCA) [9]. In the IACS- DCA, each A periodically coputes the average CCI powers (obtained fro past CCI easureents) on all available channels. The channel having the lowest average CCI power is selected. It was shown that IACS-DCA can for a stable channel reuse pattern in a distributed anner and itigate the CCI copared with the quasi-dca [1]. It is well-known that the transit power control (TC) is an effective technique to reduce the CCI [11]-[14]. TC avoids excessive transit power to reduce the CCI. The signal-tointerference power ratio (SIR) based TC was extensively studied for its application to code division ultiple access (CDMA) cellular systes [15]-[17]. However, a TC schee suitable for the IACS-DCA has not been studied yet. In this paper, we propose a SIR based TC schee suitable for IACS-DCA. We show, by coputer siulation, that the IACS-DCA with proposed SIR based TC fors a stable channel reuse pattern and can further iprove the outage probability of signal-to-interference-plus-noise power ratio (SINR). The rest of this paper is organized as follows. Sect. II describes the proposed SIR based TC suitable for the IACS- DCA. In Sect. III, we exaine by coputer siulation the stability of channel reuse pattern and the SINR achievable with the IACS-DCA using the proposed TC. Sect. IV gives soe concluding rearks. II. ROOSED SIR BASED TC We assue wireless networks with tie division duplex (TDD), in which each A is designed to periodically broadcast the beacon containing the CCI inforation (the average CCI power on the selected channel) and active obile station (STA) is designed to transit the uplink signal containing the path loss inforation in addition to uplink data. It is assued that the transit beacon power is known to all STA. STA gets the CCI inforation by receiving the beacon. Upon the reception of beacon of its A, STA easures the path loss (including shadowing loss). Using the CCI and path loss inforation, STA deterines its uplink transit power so that the received signal at its A eets the required SIR. Meanwhile, A receives the uplink signal and gets the path loss inforation of corresponding active STA. The path loss inforation is used to control the downlink transit power so that the received signal at its corresponding STA eets the

2 required SIR. A. CCI table update and channel assignent [9] Fig. 1 shows a flowchart of the proposed SIR based TC. As shown in Fig. 1(a), each A 1-1) easures the uplink instantaneous CCI powers fro other STAs on all available channels, 1-2) coputes the oving average CCI powers using past CCI easureent results, 1-3) updates the CCI table, 1-4) chooses the channel having the lowest oving average CCI power for use, and 1-5) broadcasts the beacon containing the oving average CCI power. Each A 1-6) periodically repeats the procedure of 1-1) 1-5). To copute the oving average CCI powers for all available channels, the first order filtering [18] is used. The nuber of available channels is denoted by N ch. The oving average CCI power of the -th A (A ) on the ch-th channel (ch=n ch 1) at tieslot t is given as I A = (1 β) I = (1 β) A t i= + β I i β I A A ( t i) ( t 1), (1) where IA ( t) and β ( β < 1) are the instantaneous CCI power at tieslot t and the filter forgetting factor, respectively. A looks up the CCI table and chooses the ch use -th channel having the lowest oving average CCI power as STA 2-5) periodically repeats the procedure of 2-1) 2-4). When 2-6) the uplink transission is required, STA 2-7) selects the ch use -th channel and 2-8) coputes the uplink transit power. Using the oving average CCI power and the path loss, STA deterines its uplink transit power so that the received signal at its A eets the required SIR. The uplink transit power of STA is given as =, (5) t,sta Λ I t A STA, A where Λ is the target SIR value. STA 2-9) specifies the coputed path loss STA, A in addition to uplink data and 2-1) transits the uplink signal. C. Downlink TC Meanwhile, A 3-1) receives the uplink signal and 3-2) acquires the path loss (including shadowing loss) inforation fro the corresponding active STA. When 3-3) the downlink transission is required, A 3-4) selects the ch use -th channel and 3-5) coputes the downlink transit power. Siilarly to the uplink, A deterines its downlink transit power so that the received signal at corresponding STA eets the required SIR. Then, A 3-6) transits the downlink signal. The downlink transit power is given as =. (6) t,a Λ I t A STA, A ch use = arg in { IA ( t)}. (2) ch [, N ch 1] The averaging interval of the first order filtering is given as 1/(1 β) tieslots. If a too sall β is used, averaging is not enough and the easured average CCI power varies like a variation of instantaneous CCI power; hence, β 1 is required [1]. A broadcasts the beacon containing the oving average CCI power of ch use -th channel, which is given as A START 1-6) 1-1) CCI pow. easureent 1-2) Ave. CCI pow. coputation 1-3) 1-4) 1-5) CCI table update Channel selection Beacon broadcast 3-1) Uplink signal reception 3-2) ath loss acquisition 3-3) Downlink transission request 3-4) Channel selection 3-5) Transit pow. coputation 3-6) Downlink transission END C I A = in { I }. (3) ch [, Nch 1] A B. Uplink TC As shown in Fig. 1(b), STA 2-1) receives a beacon, 2-2) gets the average CCI power on the received channel, 2-3) easures the received beacon power, and 2-4) coputes the path loss (including shadowing loss). The coputed path loss between A and -th STA (STA ) is given as where STA,A t / t and =, (4) r, STA r, STA are the transit beacon power and received beacon power easured by STA, respectively. We assue that the transit beacon power is known to all STA. (a) A B 2-6) 2-5) Uplink transission request 2-1) 2-7) Beacon reception Channel selection 2-2) 2-8) Ave. CCI pow. acquisition Transit pow. coputation 2-3) 2-9) ath loss inforation Received pow. specifics easureent 2-4) 2-1) ath loss coputation Uplink transission (b) STA Fig. 1. Flowchart of proposed SIR based TC.

3 III. COMUTER SIMUATION Coputer siulation was done to exaine the stability of channel reuse pattern of proposed schee and to evaluate the perforance of SINR distribution in the network. Table 1 suarizes the siulation condition. STAs are assued to be stationary. A easures the uplink instantaneous CCI power fro other STAs and updates the CCI table and the channel at every tieslot. A periodically broadcasts the beacon containing the oving average CCI power on the selected channel and STA transits the uplink signal containing the path loss inforation with the proposed TC. The perfect easureent of the instantaneous CCI power on each A is assued. We consider the TDD syste using orthogonal frequency division ultiplexing (OFDM) [2]. All STAs transit packets with a probability of p=1 for uplink and all As transit packet with p=1 for downlink. l τ l deviation σ, respectively. h STA ',A and STA ',A are the coplex-valued path gain with E[ 1 ( l ) 2 h = STA,A ]=1 and l ' the tie delay of the l-th path between STA and A, respectively. Cell of interest Cell Syste Channel CS-DCA TABE I. COMUTER SIMUATION CONDITIONS No. of co-channel cells N all=1 No. of channels N ch=4 No. of STAs per cell U=1 Transission probability p=1. Target SINR 2 Fading type Frequency-selective block Rayleigh ower delay profile =16-path unifor ath loss exponent α=3.5 Shadowing loss standard deviation σ=. (db) Forgetting factor of first order filtering β=.99 CCI power easureent Ideal STA R (a) Cellular structure STA, A A R A /2 (b) Geoetry of A and STA A. Network odel Fig. 2 illustrates the network odel. N all =1 cells are considered and N int. =36 cells (shadowed region in Fig. 2(a)) are the cells of interest to exaine the channel reuse pattern and SIR distribution. As shown in Fig. 2(b), an A equipped with single antenna is located at the center of each cell and one STA is located randoly in the cell. The distance between adjacent As is denoted by R A. B. ropagation channel odel We assue a frequency-selective block Rayleigh fading channel which is coposed of distinct paths. The channel ipulse response between A and STA is given by with 1 STA,A (τ) = hsta,a δ(τ τsta,a ) ' ' ' l= h (7) h STA ',A = R α STA ',A 1 ηsta',a 1 h ( l ) STA ',A, (8) where RSTA ', A, α, and η STA, A denote the distance between the STA and the A, the path-loss exponent, and the shadowing loss in db having zero-ean and standard Fig. 2. Network odel. C. Uplink odel At the transitter (STA), the binary inforation sequence is data-odulated and then, the data-odulated sybol sequence is divided into a sequence of blocks of N c sybols each. Then, N c -point inverse discrete Fourier transfor (IDFT) is applied to for the OFDM signal block. The last N g saples in each block are copied and inserted as a cyclic prefix (C) into the beginning of the signal block before transission. The transitted OFDM signal block passes through a frequency-selective fading channel. At the receiver (A), after C reoval, the received signal block is decoposed by N c - point DFT into the orthogonal subcarrier coponents. The frequency doain received signal on k-th subcarrier is expressed as Y A = 2 t,sta H STA,A x + I A STA + N A, where HSTA,A and N A k represent the channel transfer function between STA and A and the noise coponent on the k-th subcarrier, respectively. x is STA (9)

4 the data sybol transitted on the k-th subcarrier. IA is the CCI coponent which A receives and expressed as where I = 2 H x, (1) A U A t,stau { u U A : u } STAu,A STAu {,1,,N all 1} is a set of STA nubers whose channel is the sae to the A. Then, the instantaneous CCI power of ch-th channel on A at tieslot t is given by 1 I. (11) Nc 1 2 A, = A ch t I k N c k = In this paper, ideal easureent of the instantaneous CCI power is assued. D. Channel reuse pattern Fig. 3 shows a one-shot observation of channel reuse pattern fored by the IACS-DCA with proposed SIR based TC. STA location is shown in Fig. 3(a). The initial channel reuse pattern at tie t= was generated by assigning channel # to all the As (see Fig. 3(b)). It can be seen fro Fig. 3 that the IACS-DCA with proposed SIR based TC can for a stable channel reuse pattern in a distributed anner. (a) STA location (b) t= t =17 t =18 t =19 t = (c) t =17 t =18 t =19 t = (d) Fig. 3. An exaple of channel reuse pattern with the IACS-DCA using proposed SIR based TC. E. SINR distribution Fig. 4 plots the probability density function (DF) of the SINR when using the IACS-DCA with proposed SIR based TC. The SINR level was easured when t=2. It can be seen fro Fig. 4 that the SINR is distributed near the target value. Table 2 shows the root ean square (RMS) error of easured SINR. It can be seen fro Table 2 that the proposed TC can keep the SINR close to the target value. Fig. 5 plots the cuulative distribution function (CDF) of the SINR when using the proposed TC. For coparison, the SINR distributions of the quasi-dca and the IACS-DCA without TC under a CCI liited environent are also plotted. It can be seen fro Fig. 5 that the proposed TC iproves the outage probability of SINR by increasing the TC target. For exaple, the proposed TC using the target SIR (db) iproves the SINR at CDF=1 1 by about 4dB and 1dB for uplink and downlink, respectively, copared to the IACS- DCA without TC. Target value TABE II. RMS ERROR OF MEASURED SINR RMS error (db) Uplink Downlink DF DF (a) Uplink w/ TC w/ TC (b) Downlink Fig. 4. DF of SINR.

5 rob. [SINR<Abscissa] rob. [SINR<Abscissa] 1.E+ 1.E-1 Λ=16 Λ=17 Λ=18 Λ=19 Quasi-DCA w/o TC 1.E E+ 1.E-1 1.E-2 Quasi-DCA (a) Uplink Λ=16 Λ=17 Λ=18 Λ=19 w/o TC w TC (b) Downlink Fig. 5. CDF of SINR. w TC IV. CONCUSION In this paper, we proposed a SIR based TC schee suitable for IACS-DCA. We showed, by coputer siulation, that even if SIR based TC is used, the IACS-DCA fors a stable channel reuse pattern in a distributed anner. It was also shown that the proposed TC using the target SIR (db) iproves the SINR at CDF=1 1 by about 4dB and 1dB for uplink and downlink, respectively, copared to the IACS- DCA without TC under a CCI liited environent. REFERENCES [1] I. Katzela and M. Naghshineh, Channel assignent schees for cellular obile telecounication systes: a coprehensive survey, IEEE ersonal Coun., Vol. 3, No. 3, pp. 1-31, Jun [2] A. Goldsith, Wireless Counications, Cabridge Uni-versity ress, Aug 25. [3] H. Skalli, S. Ghosh, S. K. Das,. enzini, and M. Conti, Channel assignent strategies for ultiradio wireless esh networks: issues and solutions, IEEE Coun. Magazine, Vol. 45, No. 11, pp , Nov 27. [4] D. Goodan, S. A. Grandhi, R. Vijayan, Distributed dy-naic channel assignent schees, roc. IEEE 43th Ve-hicular Technology Conference (VTC1993-Spring), May [5] G. Cao and M. Singhal, Distributed fault-tolerant channel assignent for cellular networks, IEEE Journal on Selected Areas In Coun., Vol. 18, No. 7, pp , July 2. [6] H. uo and N. K. Shankaranarayanan, A distributed dy-naic channel allocation technique for throughput i-proveent in a dense WAN environent, IEEE Interna-tional Conference on Acoustics, Speech, and Signal ro-cessing (ICASS24), Vol.5, pp , May 24. [7] Y. Furuya and Y. Akaiwa, Channel segregation, a distrib-uted adaptive channel assignent schee for obile co-unication systes, IEICE Trans. Coun., Vol. E74-B, No. 6, pp , June [8] Y. Akaiwa and H. Andoh, Channel segregation-a self-organized dynaic channel allocation ethod: application to TDMA/FDMA icrocellular syste, IEEE Journal on Selected Areas in Coun., Vol. 11, No. 6, pp , Aug [9] R. Matsukawa, T. Obara, and F. Adachi, A dynaic chan-nel assignent schee for distributed antenna networks, roc. IEEE 75th Vehicular Technology Conference (VTC212-Spring), May 212. [1] Y. Matsuura, S. Kuagai, T. Obara, T. Yaaoto, and F. Adachi, Channel Segregation Based Dynaic Channel Assignent for WAN, IEEE The 13th International Conference on Counication Systes (ICCS212), Nov 212. [11] A. F. Molisch, Wireless Counications, 2nd Edition, IEEE Wiley, Nov 21. [12] M. Krunz, A. Muqattash, and S. J. ee, Transission power control in wireless ad hoc networks: challenges, solutions and open issues, IEEE Network, Vol. 18, No. 5, pp. 8-14, Sep-Oct 212. [13] D. Qiao, S. Choi, A. Jain, and K. G. Shin, Adaptive transit power control in IEEE 82.11a wireless ANs, roc. IEEE 57th Ve-hicular Technology Conference (VTC23-Spring), Aprl 23. [14] J. Zander, erforance of optiu transitter power control in cellular radio systes, IEEE Trans. on Vehicular Technology, Vol. 41, No. 1, pp , Feb [15] S. Ariyavisitakul, Signal and interference statistics of a CDMA syste with feedback power control-art II, IEEE Trans. Coun., Vol. 42, pp , Feb/Mar/Apr [16] S. Seo, T. Dohi, and F. Adachi, SIR-Based Transit ower Control of Reverse ink for Coherent DS-CDMA Mobile Radio, IEEE Trans. on Coun., Vol. E81-B, pp , July [17] K. Higuchi, H. Andoh, K. Okawa, M. Sawahashi, and F. Adachi, Experiental evaluation of cobined effect of coherent RAKE cobining and SIR-based fast transit power control for reverse link of DS-CDMA obile radio, IEEE Journal on Selected Areas in Coun., Vol. 18, No. 8, pp , Aug 2. [18] S. W. Roberts, Control chart tests based on geoetric oving averages, Technoetrics, Vol. 1, No. 3, pp , Aug [19] R. Jain, D. Chiu, and W. Hawe, A quantitative easure of fairness and discriination for resource allocation in shared coputer syste, DEC Technical Report 31, Sept [2] R. V. Nee and R. rasad, OFDM for Wireless Multiedia Counications, Artech House, 2.