A Distributed Channel Probing Schemefor Wireless NetworksChenxi Zhu and M. Scott CorsonInstitute for Systems ResearchUniversity of MarylandCollege Park, Maryland 20742email: czhu,corson(li?isr. umd.eduAbstract—Thispaper presents a dmtributed channel probing scheme forwireless networks. By transmitting a probing signal in a channel and mea-suring the signal-to-interference ratio(SIR), a wirelessnoderan estimatechanneladmissibility and predict its required transmissionpower withoutfully powering up.The channel probing scheme can be used as part of ad~tributed channel allocation algorithm, and simulations have shown thatit outperforms other schemes.ICeyword$-Wkeless networks, power control, channel probing, dynamicchannel allocation, admission control.I. INTRODUCTIONPower control and dynamic channel allocation are two effec-tive means to improve the capacity of a wireless network [1],[2], [3], [4], [5]. When trying to combine the two together,one is faced with the problem of how to characterize channelutilization and how an algorithm, running from the perspectiveof an individual node independently, can use such informationto facilitate its channel selection. Most DCA schemes use in-terference power as a criterium for channel selection [6], [7],[8]. This paper introduces a channel probing scheme which al-lows a transmitter, in co-operation with its corresponding re-ceiver, to probe a channel, to estimate the channel’s conditionand to further predict the required transmission power to meetits desired SIR. It is a fully distributed scheme which requiresno communication between different links (a link is a transmit-ter/receiver pair), yet the local admissibility of each link and theglobal feasibility of the entire system are shown to be equiva-lent. By probing channels, the algorithm executing for a givenlink can choose the best channel. Hence it can be used as partof a dynamic channel allocation scheme, which uses transmis-sion power as the criterion for channel selection. This scheme iscompared with other channel allocation schemes via simulation.The simulation results show that with the new scheme, newly-arrived transmissions experience less blocking and the on-goingtransmissions suffer less disruptions.II. THE SYSTEM MODELWe consider a TDMA (or FDMA)-based wireless networkwhere the transmitters can adjust their transmission power con-tinuously within a given range. Each time slot in a TDMA timeframe (or a carrier frequency in a FDMA system) is referredto as a channel. Nodes perform a closed loop power controlalgorithm described as follows. The power control algorithmused is the same as that in [5]. Suppose that there are M activelinks, labeled 1 through M, in a given channel. Each link con-sists of a transmitter and a receiver, and has a target signal-tointerference ratio ~t. We assume this transmitter/receiver pairis determined by some other schemes, and it is considered fixedin this paper. The terms link and transmitter/receiver pairare used interchangeably, and transmission power and SIR of alink respectively means the output power of the transmitter andthe SIR at the receiver. Let gi,j be the propagation gain betweenthe jth transmitter and the ith receiver, and let G = [gi,j] bethe transmission gain matrix of the system. The SIR of a linkis determined by the transmission powers of the active links,the transmission gain, the target SIR and the noise n; at the re-ceivers. When inter-channel interference is neglected, the SIRof link i is given by:9i,iPi-f~ =Pi——,(1)ni +x~=l,j#~ 9i,jPj‘Vi + ~~=~ zi,jpjwhere pj >0 k the transmission power of link j. The quantities,zi,j and vi are the normalized transmission gain and receivernoise, defined asTransmission power control is applied as to make sure thatthe SIR ~i of every link vi > yt, for i = 1,2, ...M. Based on itsSIR, each link updates its transmission power as,tpi(~ + 1) = mh(~pi(k),p~az), i = 1,2, ...M.(2)where pmax is the maximal transmission power of the trans-mitter. When the maximal power pmaz is not a constraint, thepower control algorithm will converges to a unique solutionP* = (1 – ‘$z)-yv, (3)if and only if the Perron eigenvalue (the largest eigenvalue) ofmatrix Z = [~i,i], pP(Z), satisfies pP(Z) < .$ [9]. The Mlinks are called admissible if they can all achieve their targetSIRS, and inadmissible otherwise. In the latter case the system iscalled’ interference-limited’, because the interference cannot beovercome simply by increasing the transmission power. Whenthe maximal transmission power is taken into consideration, itis also necessary thatP* < pm.% 1,(4)where 1 is the all 1 vector with appropriate length. If pp (Z) <$ but transmitters do not have enough power, the system iscalled ‘power limited’. Such a system can be made admissibleby increasing the maximal transmission power constraint.0-7803-7018-8/01/$10.00 (C) 2001 IEEE IEEE INFOCOM 2001III. THE CHANNEL PROBING ALGORITHMThe proposed channel probing mechanism is based on the factthat the set of active links update their transmission power con-stantly, and will react to increased interference in the channelby increasing their own power levels. When a set of new linksjoin the channel and start to transmit, these active links expe-rience additional interference, and as a consequence, will raisetheir powers accordingly. Their power increase is proportionalto the power of the new links. If the new links transmit their sig-nals at predefine power level and measure the correspondingSIR, it can estimate the channel condition. This is called ‘chan-nel probing’. These new links, by probing a channel, can predictwhether the channel is admissible and, if the answer is yes, whatis the required transmission power. To simplify the analysis, weignore the maximal power constraint in the next two sections,and assume the transmitters always have enough power. The ef-fect of limited pm.. will be discussed in Section V. The detailsof the channel probing algorithm is given as follows.Suppose a set of ill links, 1 to Al, are already transmittingin a channel, and they apply power control and have achievedtheir SIR balance with target SIR 7$. Their transmission powervector is given byPM = (1 – #z~)-l(vM + EM),(5)where PM =k,P2, .... PM]’ is their transmission power vec-tor> ZM ={zi,j }{1,...,M] x {1,...,M} is the interference matrixassociated with the M links, VM = [vl, V2, ..., vM]’ is their(thermal) noise vector, and 13M is an extraneous noise vec-tor
View Full Document
Unlocking...