楊永鵬 楊真真
摘 要:如何根據(jù)無(wú)線環(huán)境的變化實(shí)時(shí)選擇一個(gè)可靠的發(fā)送速率,進(jìn)而保證數(shù)據(jù)傳輸?shù)目煽啃允菬o(wú)線通信研究的熱點(diǎn)。針對(duì)傳統(tǒng)累積和平均法(Cumulative Sum Average,CUSUMA)由于計(jì)算機(jī)存儲(chǔ)數(shù)據(jù)位數(shù)的有限性可能導(dǎo)致的數(shù)據(jù)值超過(guò)計(jì)算機(jī)最大數(shù)據(jù)類型表示范圍,進(jìn)而造成統(tǒng)計(jì)數(shù)據(jù)丟失的問(wèn)題,提出基于指數(shù)加權(quán)移動(dòng)平均(Exponentially Weighted Moving Average, EWMA)算法,計(jì)算對(duì)應(yīng)速率下的數(shù)據(jù)發(fā)送成功率,并根據(jù)計(jì)算出的成功率實(shí)現(xiàn)動(dòng)態(tài)速率選擇。該算法可用于解決通用平均值算法導(dǎo)致的因數(shù)據(jù)過(guò)大造成的統(tǒng)計(jì)數(shù)據(jù)丟失問(wèn)題。
關(guān)鍵詞:指數(shù)加權(quán)移動(dòng)平均;無(wú)線通信網(wǎng)絡(luò)協(xié)議;速率;累積和平均法;均方誤差
DOI:10. 11907/rjdk. 182663
中圖分類號(hào):TP393
文獻(xiàn)標(biāo)識(shí)碼:A文章編號(hào):1672-7800(2019)006-0192-04
Abstract: According to the change of the wireless environment, how to select a reliable transmission rate to ensure the reliability of data transmission in real time is a hot topic in current research. Due to the limited number of bits in computer storage data, the Cumulative Sum Average (CUSUMA) method may cause the statistical data to exceed the maximum data range of the computer, resulting in the loss of statistical data. In this paper, the exponential weighted moving average algorithm (EWMA) is used to calculate the success rate of data transmission. At the same time, the EWMA algorithm is used to solve the loss of statistical data during data storage processing in the Cumulative Sum Average (CUSUMA) method.
Key Words: Exponentially Weighted Moving Average; Wireless communication network protocol; Rate; Cumulative Sum Average; Mean square error
0 引言
信息時(shí)代基于通信技術(shù)的網(wǎng)絡(luò)技術(shù)[1]已成為人們生活不可或缺的一部分?;贗EEE 802.11標(biāo)準(zhǔn)的WiFi設(shè)備是數(shù)據(jù)通信的主流設(shè)備,數(shù)據(jù)發(fā)送速率可達(dá)到數(shù)百兆乃至上千兆。無(wú)線網(wǎng)絡(luò)技術(shù)[2]憑借成本低、易配置、可擴(kuò)展性、移動(dòng)性和網(wǎng)絡(luò)架構(gòu)靈活等特點(diǎn)成為研究熱點(diǎn),廣泛應(yīng)用于智能機(jī)器人[3]、醫(yī)療保健、生物醫(yī)學(xué) [4]和工業(yè)控制自動(dòng)化[5]等領(lǐng)域。IEEE 802.11工作組先后制定了IEEE 802.11a、b、g、n、ac、ax等標(biāo)準(zhǔn),近年來(lái)又提出了基于電視未使用空白頻段的IEEE 802.11af無(wú)線網(wǎng)絡(luò)通訊協(xié)議標(biāo)準(zhǔn)。
基于IEEE 802.11協(xié)議標(biāo)準(zhǔn)的無(wú)線網(wǎng)絡(luò)由于其靈活性、簡(jiǎn)單性、速度多樣性和快速性等特點(diǎn)得到廣泛應(yīng)用,其中速率多樣性能保證在無(wú)線環(huán)境變化的情況下選擇一個(gè)合適的速率,進(jìn)而保證數(shù)據(jù)傳輸?shù)目煽啃?。如何根?jù)無(wú)線環(huán)境變化實(shí)時(shí)選擇一個(gè)可靠的速率是研究的關(guān)鍵?;谡{(diào)制方式、碼率、長(zhǎng)前導(dǎo)、短前導(dǎo)、空間流個(gè)數(shù)、長(zhǎng)間隔和短間隔組合,IEEE 802.11標(biāo)準(zhǔn)規(guī)定了多種數(shù)據(jù)傳輸速率。IEEE 802.11標(biāo)準(zhǔn)由最初傳統(tǒng)的1M、2M、5.5M、6M速率發(fā)展到現(xiàn)在MCS0、MCS1等高吞吐量速率。理論上,使用IEEE 802.11標(biāo)準(zhǔn)中的最高傳輸速率對(duì)無(wú)線數(shù)據(jù)進(jìn)行傳輸,數(shù)據(jù)傳輸?shù)耐掏铝繉⑦_(dá)到最佳。但是,由于用于傳輸無(wú)線數(shù)據(jù)的信道存在眾多電磁波干擾(比如雷達(dá)信號(hào)或其它無(wú)線設(shè)備發(fā)出的無(wú)線信號(hào)等),并且這種干擾不可控,導(dǎo)致數(shù)據(jù)傳輸環(huán)境惡劣,這種情況下選擇高速率進(jìn)行無(wú)線信號(hào)傳輸會(huì)使數(shù)據(jù)傳輸?shù)腻e(cuò)誤率和丟包率增加,不利于數(shù)據(jù)傳輸?shù)姆€(wěn)定性和正確性。
針對(duì)該問(wèn)題,無(wú)線網(wǎng)絡(luò)傳輸系統(tǒng)需要引入一種動(dòng)態(tài)速率選擇算法,該算法可以根據(jù)周邊無(wú)線信道環(huán)境實(shí)時(shí)動(dòng)態(tài)選擇合適的速率。目前,基于IEEE 802.11無(wú)線網(wǎng)絡(luò)標(biāo)準(zhǔn)的速率自適應(yīng)算法主要有兩種:①基于信道直接測(cè)量的方法,例如基于接收端的動(dòng)態(tài)速率選擇算法[6-7](Receiver-Based Auto Rate,RBAR)。該算法使用接收端估計(jì)當(dāng)前信道質(zhì)量,并通過(guò)修改RTS/CTS幀實(shí)現(xiàn)速率信息的交互。移動(dòng)環(huán)境的速率自適應(yīng)算法[8](Rate Adaptation in Mobile environments,RAM),通過(guò)控制反饋幀速率決定發(fā)送側(cè)是否改變傳輸速率。該算法雖然無(wú)需改變協(xié)議,但無(wú)法實(shí)時(shí)改變速率;②基于數(shù)據(jù)發(fā)送成功率統(tǒng)計(jì)的方法,即統(tǒng)計(jì)一段時(shí)間內(nèi)的吞吐量,從而判斷信道的傳輸質(zhì)量。例如自適應(yīng)動(dòng)態(tài)速率反饋算法[9](Adaptive Auto Rate Feedback,AARF)統(tǒng)計(jì)發(fā)送連續(xù)成功或失敗的幀個(gè)數(shù),ONOE算法[10](Only Openly Available Bit-rate Selection Algorithm,ONOE)維持當(dāng)前傳輸速率的信用度,SampleRate算法[11]統(tǒng)計(jì)各速率下數(shù)據(jù)幀的平均傳輸時(shí)間,Minstrel算法[12]則統(tǒng)計(jì)各速率對(duì)應(yīng)的傳輸吞吐率。第1種方法主要依賴于硬件,導(dǎo)致設(shè)備成本較高,另外需要修改協(xié)議,導(dǎo)致協(xié)議兼容性有一定的局限性。因此,基于統(tǒng)計(jì)信息的方法成為當(dāng)前動(dòng)態(tài)選擇速率的首選。本文主要通過(guò)統(tǒng)計(jì)各速率對(duì)應(yīng)的傳輸成功率進(jìn)行動(dòng)態(tài)速率選擇,在IEEE 802.11協(xié)議[13]中沒(méi)有一個(gè)成功率計(jì)算的統(tǒng)一標(biāo)準(zhǔn)。傳統(tǒng)的累積和平均法(Cumulative Sum Average,CUSUMA)[14]可通過(guò)記錄設(shè)備長(zhǎng)時(shí)間運(yùn)行過(guò)程中的發(fā)送成功次數(shù)和總的發(fā)送次數(shù)計(jì)算出發(fā)送成功率,該方法雖然簡(jiǎn)單,但由于計(jì)算機(jī)存儲(chǔ)數(shù)據(jù)位數(shù)的有限性,可能導(dǎo)致數(shù)據(jù)值超過(guò)計(jì)算機(jī)最大數(shù)據(jù)類型表示范圍,造成統(tǒng)計(jì)數(shù)據(jù)丟失。
本文針對(duì)計(jì)算機(jī)存儲(chǔ)數(shù)據(jù)位數(shù)的有限性,基于成功率統(tǒng)計(jì)方法,提出指數(shù)加權(quán)移動(dòng)平均(Exponentially Weighted Moving Average,EWMA)算法[15-16]。該算法可以實(shí)時(shí)統(tǒng)計(jì)不同速率下無(wú)線數(shù)據(jù)發(fā)送成功率,解決平均值算法引起的統(tǒng)計(jì)數(shù)據(jù)丟失問(wèn)題。
1 EWMA原理及設(shè)計(jì)方法
工業(yè)領(lǐng)域尤其是在數(shù)據(jù)通信領(lǐng)域,需要實(shí)時(shí)統(tǒng)計(jì)數(shù)據(jù)發(fā)送的成功率,并根據(jù)統(tǒng)計(jì)的成功率作出相應(yīng)決策,如在無(wú)線數(shù)據(jù)通信領(lǐng)域中決策發(fā)送速率[17]的選取。傳統(tǒng)統(tǒng)計(jì)成功率的方法為累積和平均法(CUSUMA),該方法首先將發(fā)送成功的數(shù)據(jù)包個(gè)數(shù)累計(jì)求和,然后統(tǒng)計(jì)總共發(fā)送數(shù)據(jù)包的個(gè)數(shù),最后將兩個(gè)值的比值作為數(shù)據(jù)發(fā)送成功率。
從圖2可以看出,在[λ=0.75]的情況下,EWMA算法與累積計(jì)算出來(lái)的概率值差值是最小的,并且隨著時(shí)間的推移兩個(gè)值趨于重合,充分證明了EWMA算法的合理性。
假設(shè)在600ms時(shí)數(shù)據(jù)出現(xiàn)翻轉(zhuǎn),由于位寬的限制,前600ms數(shù)據(jù)丟失,則計(jì)算出來(lái)的CUSUMA值為0.11,在[λ=0.75]的情況下,EWMA的值為0.21,而實(shí)際成功率為0.29。所以,在數(shù)據(jù)出現(xiàn)丟失的情況下,EWMA算法獲取的成功率值更為準(zhǔn)確。
4 結(jié)語(yǔ)
在綜合分析IEEE 802.11 協(xié)議簇的速率動(dòng)態(tài)選擇算法之后,本文針對(duì)IEEE 802.11標(biāo)準(zhǔn)的無(wú)線網(wǎng)絡(luò)通信動(dòng)態(tài)速率選擇過(guò)程中的統(tǒng)計(jì)成功率計(jì)算方法,提出了一種基于EWMA算法的數(shù)據(jù)成功率統(tǒng)計(jì)方法。該方法克服了傳統(tǒng)CUSUMA方法對(duì)數(shù)據(jù)位寬依賴的缺陷,能夠解決由于數(shù)據(jù)位寬限制導(dǎo)致的統(tǒng)計(jì)數(shù)據(jù)丟失問(wèn)題,及由此造成的成功率誤差較大的問(wèn)題,從而滿足統(tǒng)計(jì)數(shù)據(jù)成功率和實(shí)時(shí)選取發(fā)送速率的需求,能夠很好地屏蔽由于無(wú)線環(huán)境可變性導(dǎo)致的速率不穩(wěn)定現(xiàn)象。實(shí)驗(yàn)表明,EWMA算法一方面能夠?qū)崿F(xiàn)IEEE 80211標(biāo)準(zhǔn)的無(wú)線數(shù)據(jù)發(fā)送成功率統(tǒng)計(jì),另一方面能夠解決位寬限制造成的統(tǒng)計(jì)信息丟失問(wèn)題。
參考文獻(xiàn):
[1] 覃慶國(guó),楊亞培,蔣寧,等. 依托學(xué)科優(yōu)勢(shì)構(gòu)建電子信息類創(chuàng)新人才培養(yǎng)體系[J]. 電子科技大學(xué)報(bào):社會(huì)科學(xué)版,2010,12(5):103-106.
[2] 李二兵,徐偉強(qiáng). 基于信道信息統(tǒng)計(jì)的WLAN速率自適應(yīng)算法[J]. 無(wú)線電通信技術(shù),2018,44(1):55-59.
[3] HAN S,MOK A K,MENG J, et al. Architecture of a cyberphysical avatar[C]. ACM/IEEE International Conference on Cyber-Physical Systems, IEEE, 2013:189-198.
[4] ZHANG W,ZHU X,HAN S,et al. Design of a network-based mobile gait rehabilitation system[M]. Monterey:IEEE, 2012.
[5] SONG J,HAN S,MOK A,et al. Wirelesshart: applying wireless technology in real-time industrial process control[C]. Real-Time and Embedded Technology and Applications Symposium, 2008. RTAS '08,IEEE, 2008:377-386.
[6] HOLLAND G,VAIDYA N,BAHL P. A rate-adaptive mac protocol for multi-hop wireless networks[C]. ACM International Conference on Mobile Computing and NETWORKING,2001:236-251.
[7] HITHNAWI A. An on-demand rate-adaptation mechanism for IEEE 802.11 networks[D]. Achen: RWTH Aachen University, 2011.
[8] CHEN X,GANGWAL P,QIAO D. Ram: rate adaptation in mobile environments[J]. IEEE Transactions on Mobile Computing, 2012, 11(3):464-477.
[9] LACAGE M,MANSHAEI M H,TURLETTI T. IEEE 802.11 rate adaptation:a practical approach[C]. ACM International Symposium on Modeling, Analysis and Simulation of Wireless and Mobile Systems,ACM, 2004:126-134.
[10] PEFKIANAKIS I,LEE S B,LU S. Towards mimo-aware 802.11n rate adaptation[J]. IEEE/ACM Transactions on Networking,2013, 21(3):692-705.
[11] NI B,SANTHAPURI N,GRAY C,et al. Selection of bit-rate for wireless network coding[C]. Sensor, Mesh and Ad Hoc Communications and Networks Workshops, 2008,SECON Workshops '08. IEEE Communications Society Conference on. IEEE, 2008:1-6.
[12] XIA D,HART J,F(xiàn)U Q. Evaluation of the minstrel rate adaptation algorithm in ieee 802.11g wlans[C]. IEEE International Conference on Communications. IEEE, 2013:2223-2228.
[13] 胡智群. 層次網(wǎng)絡(luò)中高效WLAN的性能分析與優(yōu)化研究[D]. 北京:北京郵電大學(xué), 2018.
[14] ABBAS,NASIR,RIAZ,et al. Mixed exponentially weighted moving average-cumulative sum charts for process monitoring[J]. Quality and Reliability Engineering International, 2013, 29(3):345-356.
[15] MURRAY N B,GABBETT T J,TOWNSHEND A D,et al. Calculating acute: chronic workload ratios using exponentially weighted moving averages provides a more sensitive indicator of injury likelihood than rolling averages[J]. British Journal of Sports Medicine, 2017, 51(9):749-755.
[16] ANDRé LUCAS,ZHANG X. Score-driven exponentially weighted moving averages and value-at-risk forecasting[J]. International Journal of Forecasting,2016,32(2):293-302.
[17] 李二兵,徐偉強(qiáng). 基于信道信息統(tǒng)計(jì)的WLAN速率自適應(yīng)算法[J]. 無(wú)線電通信技術(shù),2018, 44(1):55-59.
[18] 李二兵. 無(wú)線局域網(wǎng)中基于數(shù)據(jù)鏈路層增強(qiáng)的傳輸可靠性研究[D]. 杭州:浙江理工大學(xué), 2018.
[19] 楊國(guó)華. 802.11無(wú)線局域網(wǎng)標(biāo)準(zhǔn)與安全[J]. 計(jì)算機(jī)工程與設(shè)計(jì), 2004,25(8):1349-1352.
[20] 辛艷,梁建坤,修長(zhǎng)虹. 無(wú)線局域網(wǎng)IEEE802.11ac協(xié)議的研究及應(yīng)用[J]. 網(wǎng)絡(luò)安全技術(shù)與應(yīng)用,2016,25(9):80-81.
[21] 王峰. 基于EWMA-GARCH(1,2)模型的統(tǒng)計(jì)套利策略研究[J]. ?科技經(jīng)濟(jì)導(dǎo)刊, 2018,10(8):34-35.
[22] 王蘊(yùn)紅,劉國(guó)歲. LMS算法中穩(wěn)態(tài)均方誤差的研究[J]. 南京理工大學(xué)學(xué)報(bào),1996,20(5):469-472.
[23] 徐天河,楊元喜. 均方誤差意義下正則化解優(yōu)于最小二乘解的條件[J]. 武漢大學(xué)學(xué)報(bào):信息科學(xué)版, 2004, 29(3):223-226.
(責(zé)任編輯:杜能鋼)