PDF file

www1.i2r.a.star.edu.sg

PDF file

EE5401 Cellular Mobile Communications- Should not result in severe degradation in theperformance of the system as compared to a single userscenario.- Approaches can be broadly grouped into two categories:narrowband and wideband.Multiple Accessing Techniques : with possible conflictand conflict- free- Random access- Frequency division multiple access (FDMA)- Time division multiple access (TDMA)Institute for Infocomm Research National University of Singapore 2


EE5401 Cellular Mobile Communications- Spread spectrum multiple access (SSMA): an example isCode division multiple access (CDMA)- Space division multiple access (SDMA)Narrowband Systems- Transmission experiences nonselective fading. Thismeans that when fades occur, all of the information (i.e.the whole channel) is affected.Institute for Infocomm Research National University of Singapore 3


EE5401 Cellular Mobile Communicationsx(t)XInstitute for Infocomm Research National University of Singapore 4


EE5401 Cellular Mobile CommunicationsInterleaving + coding techniques are used1 8 15 22 29 …..2 9 16 23 30 …..3 10 17 24 31 ……4 11 18 25 32 ……5 12 19 26 33 ……6 13 20 27 34 ……7 14 21 28 35 ……Data generated in source 1 2 3 4 5 … 8 9 10 11 12 … 15 16 17 18 …Transmitted data 1 8 15 22 29 … 2 9 16 23 … 3 10 17 24 …Received data 1 8 15 22 29 … 2 9 16 23 30 … 3 10 17 24 31 …with the shaded data bits undergo deep fadeDeinterleaved data 1 2 3 4 5 … 8 9 … 15 16 17 18 …Institute for Infocomm Research National University of Singapore 5


EE5401 Cellular Mobile Communications- Channelized system: generally total spectrum is dividedinto a number of relatively narrow radio channels (e.g.FDMA). Occurrence of call blocking if channels are allbeing used. Unused bandwidth in each channel cannotbe used by other users.Wideband Systems- The main feature of wideband systems is that either allthe spectrum available (e.g. CDMA, TDMA) or aconsiderable portion of it is used by each user (e.g.TDMA+FDMA).Institute for Infocomm Research National University of Singapore 6


EE5401 Cellular Mobile Communications- The advantage of wideband systems is that thetransmission bandwidth always exceeds the coherencebandwidth for which the signal experiences onlyselective fading. That is, only a small fraction of thefrequencies composing the signal is affected by fading.- Signal can be distorted and therefore equalization isneeded but unlikely that a total signal fade will occur.For voice or data communications, must assure two-waycommunication (duplexing, it is possible to talk andlisten simultaneously). Duplexing may be done usingfrequency or time domain techniques.Institute for Infocomm Research National University of Singapore 7


EE5401 Cellular Mobile Communications1. Forward (downlink) band provides traffic from the BSto the mobile2. Reverse (uplink) band provides traffic from themobile to the BS.Frequency division duplexing (FDD)- Provides two distinct bands of frequencies for everyuser, one for downlink and one for uplink.- A large interval between these frequency bands must beallowed so that interference is minimized.Institute for Infocomm Research National University of Singapore 8


EE5401 Cellular Mobile CommunicationsTime division duplexing (TDD)- In TDD communications, both directions oftransmission use one contiguous frequency allocation,but two separate time slots to provide both a forwardand reverse link.- Because transmission from mobile to BS and from BSto mobile alternates in time, this scheme is also knownas “ping pong”.- As a consequence of the use of the same frequencyband, the commuication quality in both directions isthe same. This is different from FDD.Institute for Infocomm Research National University of Singapore 9


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 10


EE5401 Cellular Mobile CommunicationsRandom AccessAloha systems- Aloha is a packet-switching system. The time intervalrequired to transmit one packet is called a slot.- When transmissions from two or more users overlap,they destroy each other, regardless of whether they arecompletely or partially overlap – collision takes place.- The maximum interval over which two packets canoverlap and destroy each other is called the vulnerableperiod.Institute for Infocomm Research National University of Singapore 11


EE5401 Cellular Mobile Communications- The mode of random access in which users can transmitat anytime is called pure Aloha. In this case, thevulnerable period is two slot times.Institute for Infocomm Research National University of Singapore 12


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 13


EE5401 Cellular Mobile Communications- A version in which users are restricted to transmit onlyfrom the instant corresponding to the slot boundary isreferred to as Slotted Aloha. The alignment oftransmission to coincide with the slot boundary meansthat packets can only experience complete overlap, sothe vulnerable period is only one slot time.Throughput of Aloha systems- In the Aloha systems, a transmitting user upon hearinga collision, backs off for a random delay interval andtransmits again, until success is achieved. This is knownas collision resolution.Institute for Infocomm Research National University of Singapore 14


EE5401 Cellular Mobile Communications- We assume that the total traffic generated fortransmission (including re-transmission) obeys a Poissondistribution.- The transmission is successful when there is no otherpacket transmission during the vulnerable period.Institute for Infocomm Research National University of Singapore 15


EE5401 Cellular Mobile Communicationsp aNewpacketsGContendingin the slotSSuccessCollidedpacketsP rCollisionInstitute for Infocomm Research National University of Singapore 16


EE5401 Cellular Mobile CommunicationsP ( success)=P(no other packet transmission occurs within a vulnerable period)- Let S be the throughput, defined as the successfullytransmitted traffic load, and G be the total offeredchannel traffic load. Assuming that both processes obeyPoisson arrivals, thenP( no other packet transmission occurs)= exp( −τG),where τ is the vulnerable period.Hence P( success ) = exp( −τG).Institute for Infocomm Research National University of Singapore 17


EE5401 Cellular Mobile Communications- By definition P ( success ) = S / G, therefore S = G exp( −τG)- We note that as S → 0 in the limit as G → ∞. Themaximum throughput for pure Aloha is given byS 1max = ≈ 0.1842(Hint: set dS= 0 )edG1whilst for slotted Aloha, S max = ≈ 0. 368 eInstitute for Infocomm Research National University of Singapore 18


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 19


EE5401 Cellular Mobile CommunicationsDelay- The delay experienced by a packet in the system ismeasured from the instant of the packet’s arrival untilthe instant the sender receives confirmation.- It is a function of the number of transmission, theretransmission delay and the time required for thesender to receive confirmation of successfultransmission.- R – number of slots to receive an acknowledgementD r – mean re-transmission delayE – mean number of transmissions until success.Institute for Infocomm Research National University of Singapore 20


EE5401 Cellular Mobile CommunicationsThe mean number of transmission until successGE = = exp( τG)SThe average packet delay for pure AlohaD = R + E −1)(R + D )( rThe average packet delay for slotted AlohaD = R + .5 + ( E −1)(⎡R⎤ + D )0 r- The mean re-transmission delay depends on thecollision resolution algorithm used.Institute for Infocomm Research National University of Singapore 21


EE5401 Cellular Mobile CommunicationstxRNACK =>retransmissionDepending on backoffl ithD r(1)RD r(2)RrxInstitute for Infocomm Research National University of Singapore 22


EE5401 Cellular Mobile CommunicationsFDMAIn FDMA, each user is allocated a unique frequency bandor channel. During the period of the call, no other usercan share the same frequency band.All channels in a cell are available to all the mobiles.Channel assignment is carried out on a first-come-firstservedbasis.The number of channels, given a frequency spectrum B T ,depends on the modulation technique (hence B w or B c)and the guard bands between the channels 2 B .guardInstitute for Infocomm Research National University of Singapore 23


EE5401 Cellular Mobile CommunicationsThese guard bands allow for imperfect filters andoscillators and can be used to minimize adjacent-channelinterference.FDMA is usually implemented in narrowband systems.Institute for Infocomm Research National University of Singapore 24


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 25


EE5401 Cellular Mobile CommunicationsB guardB wBcBcInstitute for Infocomm Research National University of Singapore 26


EE5401 Cellular Mobile CommunicationsMain features:- Continuous transmission: the channels, once assigned,are used on a non-time-sharing basis. This means thatboth subscriber and BS can use their correspondingallotted channels continuously and simultaneously.- Narrow bandwdith: Analog cellular systems use 25-30kHz. Digital FDMA systems can make use of low bitrate speech coding techniques to reduce the channelband even more.- If FDMA channels are not in use, then they sit idle andcannot be used by other users to increase capacity.Institute for Infocomm Research National University of Singapore 27


EE5401 Cellular Mobile Communications- Low ISI: Symbol time is large compared to delay spread.No equalizer is required (Delay spread is generally lessthan a few μ seconds – flat fading).- Low overhead: Carry overhead messages for control,synchronization purposes. As the allotted channels canbe used continuously, fewer bits need to be dedicatedcompared to TDMA channels.- Simple hardware at mobile unit and BS: (1) no digitalprocessing needed to combat ISI (2) ease of framing andsynchronization.Institute for Infocomm Research National University of Singapore 28


EE5401 Cellular Mobile Communications- Use of duplexer since both the transmitter and receiveroperate at the same time. This results in an increase inthe cost of mobile and BSs.AntennaDuplexerInstitute for Infocomm Research National University of Singapore 29


EE5401 Cellular Mobile Communications- FDMA required tight RF filtering to minimize adjacentchannel interference.Nonlinear effects in FDMA- In a FDMA system, many channels share the sameantenna at the BS. The power amplifiers or the powercombiners, when operated at or near saturation arenonlinear.- The nonlinearities generate intermodulation frequencies.1. Undesirable harmonics generated outside the mobileradio band cause interference to adjacent services.Institute for Infocomm Research National University of Singapore 30


EE5401 Cellular Mobile Communications2. Undesirable harmonics present inside the band of amobile cause interference to that mobile.Example: Given a nonlinear device having a transferfunctiong ( x)= k x + k x + k x12233where x is the input voltage and k I are constants.Supposing x is composed of three narrowband bandpasssignals,x =Acos a + Bcosb+ C coscInstitute for Infocomm Research National University of Singapore 31


EE5401 Cellular Mobile Communicationswith a = ( ωc − Δω)t , b = ωct, c = ( ωc + Δω)t where ω c is thecarrier frequency and Δ ω is the separation betweenadjacent carrier frequencies.g(x)= C141 cosa+ C b + C c +4442cos cos24444433wanted signal+ C144 cos(2b− a)+ C b − c + C a − b + c +444444445 cos(2 ) cos( )244444444463intermodulation productsterms consist of cos[ 2a,2b,2c,a + b,a − b,a + c,a − c,b + c,+b − c,2a− b,2a− c,2c− a,2c− b,a + b − c,−a+ b + c]1444444444442444444444443out of band, exceeding the range ω −Δω


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 33


EE5401 Cellular Mobile CommunicationsTDMATDMA systems divide the channel time into frames. Eachframe is further partitioned into time slots. In each slotonly one user is allowed to either transmit or receive.Unlike FDMA, only digital data and digital modulationmust be used.Each user occupies a cyclically repeating time slot, so achannel may be thought of as a particular time slot ofevery frame, where N time slots comprise a frame.Institute for Infocomm Research National University of Singapore 34


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore35


EE5401 Cellular Mobile CommunicationsFeatures- Multiple channels per carrier or RF channels.- Burst transmission since channels are used on a timesharingbasis. Transmitter can be turned off during idleperiods.- Narrow or wide bandwidth – depends on factors such asmodulation scheme, number of voice channels per carrierchannel.- High ISI – Higher transmission symbol rate, henceresulting in high ISI. Adaptive equaliser required.Institute for Infocomm Research National University of Singapore 36


EE5401 Cellular Mobile Communications- High framing overhead – A reasonable amount of the totaltransmitted bits must be dedicated to synchronizationpurposes, channel identification. Also guard slots arenecessary to separate users.Institute for Infocomm Research National University of Singapore 37


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 38


EE5401 Cellular Mobile Communications- The use of digital technology permits the inclusion ofseveral facilities in the mobile unit, increasing itscomplexity. One example is the use of slow frequencyhopping to counteract multipath fading.- A guard time between the two time slots must be allowedin order to avoid interference, especially in the uplinkdirection. All mobiles should synchronize with BS tominimize interference.Institute for Infocomm Research National University of Singapore 39


EE5401 Cellular Mobile CommunicationsGuard TimeBSMS1MS2T 1Tx1 Tx2 Rx1 Rx2T 2MS has to transmitSlightly earlier or someallowable guard time.Institute for Infocomm Research National University of Singapore 40


EE5401 Cellular Mobile Communications- Flexible data rates by assigning multiple time slots todifferent users based on their demand.- No duplexer used since uses different time slots fortransmission and reception. Even if FDD is used, aswitch rather than a duplexer is required.- Lower BS cost: the use of multiple channels per carrierchannel provides a proportional reduction of theequipment at the BS.- Efficient power utilization: FDMA systems require a 3- to6-dB power backoff in order to compensate forintermodulation effects.Institute for Infocomm Research National University of Singapore 41


EE5401 Cellular Mobile Communications- Efficient handoff : TDMA systems can take advantage ofthe fact that the transmitter is switched off during idletime slots to improve the handoff procedure. Anenhanced link control, such as that provided by mobileassisted handoff (MAHO) can be carried out by asubscriber by listening to neighboring BS during the idleslot of the TDMA frame.- In TDMA/TDD system, half of the time slots in theframe information message are used for forward linkchannel and half for reverse link channels.Institute for Infocomm Research National University of Singapore 42


EE5401 Cellular Mobile Communications- In TDMA/FDD systems, same frame structure can beused for both forward and reverse transmission butcarrier frequencies used are different.Efficiency of TDMA, η f : is a measure of the percentage ofbits per frame which contain transmitted data. Thetransmitted data include source and channel coding bits.ηf=bT− bbTOH⋅100%b OH includes all overhead bits such as preamble, guardbits, etc.Institute for Infocomm Research National University of Singapore 43


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore44


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 45


EE5401 Cellular Mobile CommunicationsSynchronous and asynchronous TDMA- Previously we mentioned the same slot of all the framesare assigned to the same user. The slot assignment canactually be fixed or dynamic.- If the assigned slot is fixed from frame to frame for theduration of the connection, the users have tosynchronize to their respective assigned slots. Thismode of TDMA is referred to as STDMA.- With packet-switched transmission, it is more efficientto allow a user to transmit only when it has packets toInstitute for Infocomm Research National University of Singapore 46


EE5401 Cellular Mobile Communicationssend. Transmission slots are dynamically assigned fromframe to frame. This mode is known as ATDMA.- ATDMA is implemented using a reservation accessmechanism.Institute for Infocomm Research National University of Singapore 47


EE5401 Cellular Mobile CommunicationsSSMASpread spectrum systems: The desired signal istransmitted over a bandwidth which is much larger thanthe Nyquist bandwidth. It is first developed for militaryapplications for1. Security2. Undetectability: minimum probability of beingdetected3. Robust against intentional jammersInstitute for Infocomm Research National University of Singapore 48


EE5401 Cellular Mobile CommunicationsApplications- Security- Robust against unintentional interference- It is not bandwidth efficient when used by a single userbut has the capability to overcome narrowband jammingsignals (cannot overcome AWGN or wideband jammingsignal) and multipath.- Providing multiple access- If many users can share the same spread spectrumbandwidth without interfering with one another,Institute for Infocomm Research National University of Singapore 49


EE5401 Cellular Mobile Communicationsbandwidth efficient improved but will affect the capabilityto overcome jamming.Institute for Infocomm Research National University of Singapore 50


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore51


EE5401 Cellular Mobile CommunicationsTwo major spreading techniques are available- Direct sequence- Frequency hoppingDirect sequence spreading- Data symbol is modulated by a high-rate pseudorandomsequence c c L c ], c = {−1,1 }.r= [ 0 M −1- The Nyquist bandwidth requirements for transmittingone bit is B = 1 T .bb- The bandwidth requirements for the spreadtransmission is B = 1 T .cciInstitute for Infocomm Research National University of Singapore 52


EE5401 Cellular Mobile Communicationsdata period T schip period T cc rInstitute for Infocomm Research National University of Singapore53


EE5401 Cellular Mobile Communications- The sequence c r does not necessarily span only one bitinterval. The number of chips in one bit interval isusually denoted by N.- If M = N , the sequence spans exactly one bit interval.The sequence is repeated for consecutive bits. This istermed short codes. Examples are Gold codes, Walshcodes. Short code is sensitive to multiple path effectsince it loss the orthogonity.- If M >> N , a very long sequences is generated, e.g.2 42 − 1, and segments of length N of the sequence is usedfor modulating (spreading) each data bit. This is termedlong codes and it appears as if a random modulationInstitute for Infocomm Research National University of Singapore 54


EE5401 Cellular Mobile Communicationssequence is chosen for each consecutive bit. Examplesare PN or m-sequence.- Auto- and cross correlation functions of random codesInstitute for Infocomm Research National University of Singapore 55


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 56


EE5401 Cellular Mobile CommunicationsDS-SS transmitter/receiver- ci= {−1,1 } carrier power P, T d is the transmission delaythrough distortionless path.- Assuming BPSK as the modulation technique.Institute for Infocomm Research National University of Singapore 57


EE5401 Cellular Mobile Communications2 P⋅ c( t)cos{ ωt+θ ( t)}0iBinarydataPhasemodulator2Pcos{ ωt+θ ( t)}0is () td2Pcosωt0ct ()Spreading2 P⋅c( t− T )cos{ ω t+ θ ( t− T ) + φ}d 0 i d+ Noise + InterferenceBandpassfilterData phasedemodulatorct ( −Tˆd)DespreadingEstimateddataInstitute for Infocomm Research National University of Singapore58


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 59


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 60


EE5401 Cellular Mobile CommunicationsFigure: When single jamming tone is presenceInstitute for Infocomm Research National University of Singapore 61


EE5401 Cellular Mobile CommunicationsWhen single jamming tone is presence- Observe that the data signal has been despread to thedata bandwidth, while the single-tone jammer has beenspread over the full transmission bandwidth of thespread-spectrum system.- A filtering operation therefore rejects a large fraction ofthe spread jammer power. It can be shown that this isreduced by a factor of T c / T relative to its value withoutthe use of spread spectrum.Institute for Infocomm Research National University of Singapore 62


EE5401 Cellular Mobile CommunicationsAn important factor is the processing gain which quantifiesthe system’s ability to combat interference and multipatheffects, G = T T NP c =Institute for Infocomm Research National University of Singapore 63


EE5401 Cellular Mobile CommunicationsFrequency-hop spread spectrum (FH-SS)- To change the frequency of the carrier periodically.- Typically, each carrier frequency is chosen from a set of 2 kfrequencies which are spaced at about the width of thedata modulation bandwidth.- The spreading code is used to control the sequence ofcarrier frequencies.- In the receiver, frequency hopping is removedsynchronously.Institute for Infocomm Research National University of Singapore 64


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 65


EE5401 Cellular Mobile CommunicationsTwo basic FH systems:- Slow frequency hopping (SFH) systems, in which severalsymbols of information are transmitted on eachfrequency hop.- One example is the M-ary frequency shift keying.- In the next Figure: (a) slow frequency hop, (b) withoutfrequency hopped, where T b : bit period, T s : symbolperiod, T c : frequency hop period.Institute for Infocomm Research National University of Singapore 66


EE5401 Cellular Mobile CommunicationsData 0 1 1 1 0 0 1 1 1 1 0 1 1 0 0 0 0 0 0freq1111011001100100110013 bits ofFH000tT bT sT cFH 000100101011111110001010Without frequencyhoppingf d10110100Institute for Infocomm Research National University of Singapore 67


EE5401 Cellular Mobile Communications- Fast frequency hopping (FFH) systems, in which severalhops occur during the transmission of one symbol.- In the next Figure: (a) fast frequency hop, (b) withoutfrequency hopped, where T b : bit period, T s : symbolperiod, T c : frequency hop period.Institute for Infocomm Research National University of Singapore 68


EE5401 Cellular Mobile CommunicationsData 0 1 1 1 0 0 1 1 1 1 0 1 1 0 0 0 0 0 0 1fre1111011001100100110013 bits ofFH code000f dTb= TT scFH code 000100101011111110001010Without frequency hoppingGray coded 4FSK10110100tInstitute for Infocomm Research National University of Singapore 69


EE5401 Cellular Mobile CommunicationsMultiple Access Techniques IICDMAEach user is assigned with a distinct signature sequence.The user employs to modulate and spread the informationsignal. When several users transmit at the same time, theyare distinguished by their spreading codes. In the next Figure, Asynchronous: τ1 ≠ L ≠ τ K, Synchronous:τ 1 = L = τ K.Downlink flat fading: Synchronous +(Common channel)jφ1ξ e = L =1ξKejφKInstitute for Infocomm Research National University of Singapore 70


EE5401 Cellular Mobile CommunicationsUplink flat fading:jφ1jφKξ1e ≠ L ≠ ξ K eAWGN: ξ1 = L = ξ K = transmit power of user, φ1 = L = φK= 0Institute for Infocomm Research National University of Singapore 71


EE5401 Cellular Mobile Communicationsa 1(n )1(t ) cos( 2π t + φ1)g 1MMMf c 1ξ1ejφchannelτ( )1 ts(t)To receivers)a K)(n)τ Ks K (t(t) g Kcos( 2π fct+ φKj Kξ K e φInstitute for Infocomm Research National University of Singapore 72


EE5401 Cellular Mobile CommunicationsCDMA channel model (AWGN or common channel)- Shared by K simultaneous users, each assigned asignature waveform g k (t)of duration T (symbolinterval).N= − 1n=0g ( t)∑ a ( n)p(t − nT ), 0 ≤ t ≤ T ,kkcwhere { a k ( n),0≤ n ≤ N −1}is a signature sequence (e.g.,pseudo-noise PN code, Gold code, etc) consisting of Nchips that take values {± 1}, p (t)is a pulse shape ofduration T c .Institute for Infocomm Research National University of Singapore 73


EE5401 Cellular Mobile Communications- Without loss of generality, we assume that all K signaturewaveforms have unit energy.T∫0g2k ( t)dt = 1.- The cross-correlation between pair of signature waveformsis given asρ ( τ ) =ijT∫0gi( t)⋅gj( t −τ) dt


EE5401 Cellular Mobile Communicationssk( t)=Lξ ∑b( i)g ( t − iT ), 0 ≤ t ≤ T ,ki=0kkwhere ξ k is the signal energy per bit, b k is the binary dataof user k.- The composite signals(t)=K∑Lξ ∑b( i)g ( t − iT −τ), 0 ≤ t ≤ T .kk= 1 i=0kk- Assuming (phase + symbol) synchronous transmission,τ k=0. In this case, it is sufficient to consider the signalreceived in one signal interval. Consider the case i = 1,kInstitute for Infocomm Research National University of Singapore 75


EE5401 Cellular Mobile Communicationsr(t)=K∑k=1ξ bkk(1) gk( t)+n(t)0≤t≤TInstitute for Infocomm Research National University of Singapore 76


EE5401 Cellular Mobile CommunicationsOptimum Receiver- Optimum receiver is designed to select the bit sequencebˆ which maximizes the conditional probability P( b′ r(t)).- Assuming that the transmitted bits are independent andequiprobable, this is equivalent to maximizing thelikelihood function⎪⎧T⎡K21⎤ ⎪⎫P( r(t)b ′)= C exp′⎨−∫ ⎢r(t)− ∑ ξkbk(1) gk( t)⎥dt2⎬⎪⎩ 2σ0 ⎣ k=1⎦ ⎪⎭or computes the log-likelihood function.Institute for Infocomm Research National University of Singapore 77


EE5401 Cellular Mobile CommunicationsTK⎡⎤Λ( b ′)= ∫ ⎢r(t)− ∑ ξkb′k (1) gk( t)⎥dt0⎣k = 1⎦[ ]where b ′ = b′b′L b′1 2K is the informationsequence selected. The one that gives minimum Λ (b′)will be chosen, i.e.,bˆ= arg min[ Λ(b′)] .{ }b′TK⎡⎤- Expand Λ( b ′)= ∫ ⎢r(t)− ∑ ξkb′k (1) gk( t)⎥dt , we have⎣ k=⎦0 122Institute for Infocomm Research National University of Singapore 78


EE5401 Cellular Mobile CommunicationsTKT2∫r( t)dt − 2 ∑ ξkbk0 k=10+KK′(1) ∫r(t)g∑ ∑ ξ ξ b′(1) b′(1) ∫ gj k j kj= 1 k=1 0Tjk( t)dt( t)gk( t)dtThe common term is ignored since it is of no relevance indetermining which sequence was transmitted.Since ρ (0) ∫ g ( t)g ( t)dt,jk= T j k0of correlation metricsΛ can be expressed in the formInstitute for Infocomm Research National University of Singapore 79


EE5401 Cellular Mobile CommunicationsC(r,β)= 2K∑k=1= 2βTξ bky − βkT′(1) ⋅Rβyk−K∑K∑j= 1 k=1ξjξ bkj′(1) bk′(1) ⋅ ρwhere R is the correlation matrix, with elements ρ (0),β = [ ξ b1(1)L ξ K b (1)] .1 K- Altogether 2 K possible choices of the bits in theinformation sequence of the K users. The optimumdetector computes the correlation metrics for eachsequence and makes a decision. The complexity thatgrows exponentially with the numbers of users, K.Complexity is measured in the number for arithmeticjk(0)jkInstitute for Infocomm Research National University of Singapore 80


EE5401 Cellular Mobile Communicationscomplexity, measured in the number of arithmeticoperations per modulation symbol.- Expression for asynchronous case can be similarlyobtained except it will involve two bits of the interferingusers.Suboptimum detectors- Optimum detector has a computational complexitygrows exponentially with K. So we are looking for suboptimumsolution.- E.g., decorrelator, MMSE, interference cancellers etcInstitute for Infocomm Research National University of Singapore 81


EE5401 Cellular Mobile CommunicationsConventional detector- The simplest suboptimum detector is the conventionaldetector (or single user detector).- The receiver of each user consists of a demodulator thatcorrelates (or matched filters) the received signal with thesignature sequence of the user, passes the correlatoroutput to the detector that makes a decision based onthe correlator output.Institute for Infocomm Research National University of Singapore 82


EE5401 Cellular Mobile CommunicationsT∫ b0() dtMatched filterInstitute for Infocomm Research National University of Singapore 83


EE5401 Cellular Mobile Communicationsr (t)Matchedfilter user 1Matchedfilter user 2y 1Sync 1y2∧b 1∧b 2Sync 2Matchedfilter user Ky Kb ∧ KSync KInstitute for Infocomm Research National University of Singapore84


EE5401 Cellular Mobile Communications- The output at the correlator for the kth user for the signalin the interval 0 ≤ t ≤ T is given byyk==T∫ r(t)g0ξ bkkk( t)dt(1) +K∑ ξ b (1) ρ (0) +j=1j≠kjj14444244443jkInterference+Noisewhere the noise component n (1) ∫ n(t)g ( t)dt is AWGNknk(1)= T k0Institute for Infocomm Research National University of Singapore 85


EE5401 Cellular Mobile Communicationswith PSD N / 2, the variance is0T2 1 2 N02 002[ (1) ]E nk = N ∫ gk( t)dt =- If the signature sequences are orthogonal ( ρ = 0for j ≠ k ), the interference from the other users vanishesand the conventional single-user detector is optimum.Multiple access interference (MAI)- MAI exits if the signature sequences are nonorthogonal( ρ ≠ 0 for j ≠ k ).jkjkInstitute for Infocomm Research National University of Singapore 86


EE5401 Cellular Mobile Communications- MAI is the limiting factor for a CDMA system employingthe conventional detector. The conventional detectormakes no attempt to overcome MAI and assumes thatthe aggregate noise plus MAI is white and Gaussian.- Asynchronous transmission is more vulnerable tointerference from other users. This is because it is moredifficult to design signature sequences for any pair ofusers that are orthogonal for all possible relative delay.Suitable multiuser detectors must be used.Institute for Infocomm Research National University of Singapore 87


EE5401 Cellular Mobile CommunicationsPower control- If one or more of the other signature sequences are notorthogonal to the user signature sequence, theinterference from other users can become excessive ifthe received power levels of one or more of the otherusers is sufficiently larger than the power level of the k-th user. This is known as the near-far problem inmultiuser communications.- The near-far problem leads to strict power controlrequirements for a CDMA system using the conventionaldetector.Institute for Infocomm Research National University of Singapore 88


EE5401 Cellular Mobile Communications- Power control is needed to maintain the required linkQoS by adapting to MS movements and also thestatistical variations in radio propagation paths. Thishelps to mitigating interference to increase the systemcapacity.- Without power control and simply let all MSs transmitat same power, their signals will arrive at BS with vastlydifferent power levels (near-far effect).- When an user transmits at excessive power level, itsperformance gets improve but will degrade theperformance of other MSs. Others will try to increasetheir transmit power to restore their link reliability, andInstitute for Infocomm Research National University of Singapore 89


EE5401 Cellular Mobile Communicationshence results in “unhealthy” competition among all MSsare not properly co-ordinated.- In the uplink, power control is to minimize powerconsumption and prolong the battery operating time.Institute for Infocomm Research National University of Singapore 90


EE5401 Cellular Mobile CommunicationsDecorrelatorNote that hereafter, you could assume all users aretransmitting at a constant power. The factorξ k = A k = α exp( jφ)in the received signal can therefore betreated as the complex channel gain. Here we areconsidering multiuser detectors and not single userdetector, hence near far problem is not our main concern.- For a K-user system, we can write the correlatoroutputs y k , k = 1,..K into a matrix form given byy = Rβ + n = RAb + nInstitute for Infocomm Research National University of Singapore 91


EE5401 Cellular Mobile Communicationswhere R is the cross correlation matrix whose elementsare ρ , b = b (1) L b (1) ] consists of bits to bejk[ 1 Krecovered, A = Diag[ξ1 L ξ K ].R−1 −1y=Ab- For BPSK, the detected symbolˆ −b = sgn( R1 y)+RnInstitute for Infocomm Research National University of Singapore 92


EE5401 Cellular Mobile CommunicationsMatchedfilter user1y 1Sync 1∧b1r(t)Matchedfilter user2y 2Ldec=R−1∧b2Sync 2Matchedfilter userKy Kb ∧KSync KInstitute for Infocomm Research National University of Singapore 93


EE5401 Cellular Mobile Communications- MAI is completely eliminated and less computationalcomplexity than optimum detector. Besides, noknowledge of the received amplitudes needed.- Example, when only two users⎡1ρ⎤1 1T−1⎡ − ρ⎤R = ⎢ ,, ( ) ( )1⎥ R =2 ⎢12 1 211⎥ ρ = ∫ g t g t dt⎣ρ⎦ − ρ ⎣−ρ ⎦ 0r(t)= ξ1b1g1(t)+ ξ2b2g2( t)+ n(t)TAfter the correlator r(t)g ( t)dt of 1 and 2∫0iInstitute for Infocomm Research National University of Singapore 94


EE5401 Cellular Mobile Communicationsy=⎡⎢⎣yy12⎤⎥⎦=⎢ ⎣⎡ρξ b111ξ b+ ρ1+ξ b2ξ bo −1⎡ ξ⎤1b1+ ( n1− ρn2) (1 − ρ )b = R y = ⎢2 ⎥,⎣ ξ2b2+ ( n2− ρn1) (1 − ρ ) ⎦ˆob = sign(b )2- The transformation R -1 has removed the interferencecomponents between the two users.22++nn12⎤⎥⎦2Institute for Infocomm Research National University of Singapore 95


EE5401 Cellular Mobile Communicationsr(t)T∫ b1( ) dt0−g ( )1 tT∫ b0( ) dtρy 1 ( t)+ˆbg 2(t )T∫ b0() dty 2 ( t)+−ˆb 2Institute for Infocomm Research National University of Singapore 96


EE5401 Cellular Mobile CommunicationsMMSE (Minimum mean-squared error) Detector- The conventional detector is optimized to combat thebackground white noise exclusively, whereas thedecorrelating detector eliminates the multiuserinterference disregarding the background noise. Incontrast, the MMSE linear detector can be seen as acompromise solution that takes into account the relativeimportance of each interfering user and the backgroundnoise.- MMSE detector implements the linear mapping whichminimizesInstitute for Infocomm Research National University of Singapore 97


EE5401 Cellular Mobile Communicationsε = E[2]b − L yMMSEThe solution can be shown to be,LMMSE⎛= ⎜R+⎝N20A−2⎟⎠⎞−1hencebˆ=LMMSE ⋅yInstitute for Infocomm Research National University of Singapore 98


EE5401 Cellular Mobile CommunicationsMF 1y 1Sync 1ˆb1r(t)MF 2MF 3y 2y 3Sync 2Sync 3[ R + N A2R -1 0 −2−1]ˆb2ˆb3MF Ky KSync KbˆKMMSE detector for synchronous channelInstitute for Infocomm Research National University of Singapore 99


EE5401 Cellular Mobile Communications- The MMSE detector is very similar to that of thedecorrelator. It has an improved performance in thepresence of noise. As the background noise goes to zero,MMSE detector converges to the decorrelating detector.However, the presence of noise will lead to a performancethat depends on the interfering signal powers, i.e.,channel information are required.- Example for a two users systemInstitute for Infocomm Research National University of Singapore 100


EE5401 Cellular Mobile CommunicationsLMMSE=⎛⎜⎝R+N20A−2⎞⎟⎠−1=2⎛⎜σ1 +⎝ ξ112⎞⎛ ⎟⎜σ1 +⎠⎝ξ2⎞⎟⎠− ρ2⎡⎛ ⎢⎜σ1 +⎢⎝ξ2⎢⎢ − ρ⎢⎣2⎞⎟⎠− ρ2⎛⎜σ1 +⎝ ξ1⎤⎥⎥⎞⎥⎟⎥⎠⎥⎦Institute for Infocomm Research National University of Singapore 101


EE5401 Cellular Mobile Communicationsσ1 +ξ22T∫ b0( ) dt+−ˆb 1r(t)g 1 ( t)T∫ b0( ) dtρg 2 ( t)T∫ b0( ) dt+−ˆb 2Institute for Infocomm Research National University of Singapore 102σ1 +ξ12


EE5401 Cellular Mobile CommunicationsSuccessive interference canceller – basic principle- The operation of the interference canceller is based onsuccessive cancellations of interference from thereceived waveform. It is important to cancel thestrongest detected signal before detection of the othersignals because it has the most severe effect inproducing interference and it is easier to achieveacquisition and demodulation of strong signals.Institute for Infocomm Research National University of Singapore 103


EE5401 Cellular Mobile Communications- Many examples but the principle is the same. Twosimple scheme are serial interference cancellation (SIC)and parallel interference cancellation (PIC).Institute for Infocomm Research National University of Singapore 104


EE5401 Cellular Mobile Communications• Features of CDMA- All users can share the same radio channel. Hence, nofrequency planning is necessary.- Soft capacity. CDMA systems do not experience blockingprobability of any kind. In theory, new calls are alwayspossible. The corresponding signals will be spreadthroughout the spectrum in the same way as thosesignals already established. Increasing the number ofusers increases the interference in a somewhat linearmanner. The only consequence is the gracefulperformance degradation of the signal-to-interferenceratio.Institute for Infocomm Research National University of Singapore 105


EE5401 Cellular Mobile Communications- Soft handover. Since the same radio channel is used forall base stations, the signal from several base stationscan be coherently combined for terminals at theboundary. This provides a “make-before-break”handover known as soft handover.- Multipath effects are mitigated through spreadspectrum transmission and through the use of a RAKEreceiver.Institute for Infocomm Research National University of Singapore 106


EE5401 Cellular Mobile CommunicationsSDMA- SDMA controls the radiated energy for each user inspace. It serves different users by using spot beamantennas. These different areas covered by the antennabeam may be served by the same frequency (in TDMA orCDMA) or different frequencies (in FDMA system).- Sectorized antennas (b) may be thought of as a primitiveapplication of SDMA.- An ideal adaptive antenna (c) is able to form a beam foreach user in the cell of interest, and the base stationtracks each user in the cell as it moves.Institute for Infocomm Research National University of Singapore 107


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 108


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore109


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 110


EE5401 Cellular Mobile CommunicationsInstitute for Infocomm Research National University of Singapore 111


EE5401 Cellular Mobile Communications- A marriage between antenna technology and digitaltechnology.Adaptive beamforming- An adaptive beamformer is a device that is able toseparate signals collocated in the frequency band butseparated in the spatial domain.- This provides a means for separating a desired signalfrom the interfering signals.Institute for Infocomm Research National University of Singapore 112


EE5401 Cellular Mobile CommunicationsFeatures:- A large number of independently steered high-gain beamscan be formed without any resulting degradation in SNRratio.- Beams can be assigned to individual users, therebyassuring that all links operate with maximum gain.- Adaptive beamforming can be easily implemented toimprove the system capacity by suppressing cochannelinterference.Institute for Infocomm Research National University of Singapore 113

More magazines by this user
Similar magazines