Lecture ¡3 Cellular ¡Systems I-Hsiang Wang ihwang@ntu.edu.tw 3/13, 2014
Cellular ¡Systems: ¡Additional ¡Challenges • So far: focus on point-to-point communication • In a cellular system (network), additional issues arise: Multiple access Inter-cell interference management 2
Issues ¡Less ¡Emphaized ¡in ¡the ¡Lecture • Handoff (focus of the network layer) • Duplexing between uplink and downlink: - Frequency Division Duplex (FDD) - Time Division Duplex (TDD) • Sectorization • Focus mainly on licensed cellular systems - WiFi, various wireless personal communication systems, are not discussed here 3
Some ¡History • Cellular concept (Bell Labs, early 70’s) • AMPS (analog, early 80’s) • GSM (digital, narrowband, late 80’s) • IS-95 (digital, wideband, early 90’s) • 3G/4G systems 4
Plot • Three cellular system designs as case studies to illustrate approaches to multiple access and (inter-cell) interference management • Both uplink and downlink will be mentioned Downlink Uplink 5
Outline • Narrowband (GSM) • Wideband system: CDMA (IS-95, CDMA 2000, WCDMA) • Wideband system: OFDMA (Flash OFDM, LTE) 6
Narrowband ¡Systems
Basic ¡Ideas • Total bandwidth divided into narrowband sub-channels - GSM: 25 MHz → 200 kHz × 125 sub-channels - Uplink (890 – 915 MHz) and Downlink (935 – 960 MHz): the same • Time Division Multiple Access (TDMA) - Users share time slots in a sub-channel; each user per time slot - Multiple access is orthogonal: intra-cell users never interfere with each other • Partial Frequency Reuse - Neighboring cells uses disjoint sets of sub-channels - Careful frequency planning → essential no inter-cell interference 8
Time ¡Division ¡Multiple ¡Access GSM: 8 users share a 200 kHz sub-channel, time slot: 577 μ s 125 sub-channels 25 MHz 200 kHz 577 μ s TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 8 users per sub-channel 9
Partial ¡Frequency ¡Reuse • Neighboring cells uses disjoint sets of sub-channels 7 • Each cell gets only 1/7 of the 3 1 1 6 total bandwidth 6 4 5 5 7 2 • Frequency reuse factor = 1/7 2 7 3 1 3 1 6 1 6 4 5 • High SINR, but price to pay: 4 5 2 5 2 7 3 - Reducing the available 7 3 1 degrees of freedom 1 6 4 - Higher complexity in 6 4 5 network planning in real world 10
Time-‑Frequency ¡Resource ¡Allocation Frequency user ¡index ¡ 1 2 3 4 5 6 7 8 cell 4 within ¡a ¡cell cell 3 9 10 11 12 13 14 15 16 cell 2 cell 1 Time 11
Time ¡and ¡Frequency ¡Diversity • Time diversity: Coding + Interleaving • Frequency diversity - Within a narrowband sub-channel: flat fading ⟹ no diversity - Obtained via frequency hopping Frequency 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 9 10 11 12 13 14 15 16 1 2 3 4 5 6 7 8 Time 12
Why ¡Full ¡Frequency ¡Reuse ¡won’t ¡Work SINR = | h | 2 P • Signal-to-Interference-plus-Noise Ratio N 0 + I • Limiting factor: interference power I - I is due to the single interferer from the neighbor cell - I is random since the location of the single interferer is uncertain - Variance of I is quite large and I can be comparable with | h | 2 P - Like deep fade, but can’t be handled by current diversity schemes • Interference averaging is desired: - If interference come from multiple interferers with smaller power, then a similar effect in diversity schemes will emerge due to LLN! N N becomes X X I k , E [ I ] = E [ I k ] I − − − − − → k =1 k =1 13
Summary • Orthogonal narrowband channels are assigned to users within a cell • Users in adjacent cells can’t be assigned the same channel due to lack of interference averaging across users ⟹ reduces the frequency reuse factor and leads to inefficient use of the total bandwidth • The network is decomposed into a set of high SINR point-to-point links, simplifying the physical-layer design • Frequency planning is complex, particularly when new cells have to be added 14
Wideband ¡System: ¡CDMA
Features ¡of ¡CDMA • Universal frequency reuse: - All users in all cells share the same bandwidth • Main advantages: - Maximizes the degrees of freedom usage - Allows interference averaging across many users - Soft capacity limit (i.e., no hard limit on the # of users supported) - Allows soft handoff - Simplify frequency planning • Challenges - Very tight power control to solve the near-far problem - More sophisticated coding/signal processing to extract the information of each user in a very low SINR environment 16
Design ¡Goals • Make the interference look as much like a white Gaussian noise as possible: - Spread each user’s signal using a pseudonoise sequence - Tight power control for managing interference within the cell - Averaging interference from outside the cell as well as fluctuating voice activities of users • Apply point-to-point design for each link - Extract all possible diversity in the channel 17
Point-‑to-‑Point ¡Link ¡Design • Extracting maximal diversity is the name of the game - Because each user has an equivalent point-to-point link! • Time diversity is obtained by interleaving across different coherence time periods and (convolutional/turbo) coding • Frequency diversity is obtained by the Rake receiver – combining of the multipaths • Transmit diversity is supported in 3G CDMA systems 18
CDMA ¡Uplink k � m � + ja Q k � m � s Q x k � m � = a I k � m � s I k � m �� m = 1 � 2 �� � � � user 1 Tx I user 1 Ch. { a 1 [ m ]} × I { s 1 [ m ]} h (1) + Q { w [ m ]} { a 1 [ m ]} × Q { s 1 [ m ]} BS Rx Σ I { a K [ m ]} × I { s K [ m ]} h ( K ) + Q { a K [ m ]} user K Ch. × Q { s K [ m ]} user K Tx � � K h � k � � � y � m � = ℓ � m � x k � m − ℓ� + w � m �� k = 1 ℓ 19
Statistics ¡of ¡Interference ¡(1/2) • Pseudorandom sequence properties: - Different users use different random shift of a sequence generated by maximum length shift register (MLSR): s [ G − 1] ⇤ T ⇥ s [0] s [1] · · · - I and Q channels of the same user can use the same sequence G − 1 ( - l = 0 G, X Near-orthogonal property: s [ m ] s [ m + l ] = � 1 , l 6 = 0 m =0 • Effective interference for user 1: h ( k ) X X I [ m ] := x k [ m − l ] l k> 1 - l Circular symmetric because each h l ( k ) is • Second-order statistics: approximately white ( k> 1 E c = P l = 0 h [ m ] | 2 i k , | h ( k ) | x k [ m ] | 2 ⇤ X E c ⇥ k := E E E [ I [ m ] I [ m + 1] ∗ ] l ⇡ 0 , l 6 = 0 l 20
Statistics ¡of ¡Interference ¡(2/2) • Due to central limit theorem (CLT), further approximate the interference as a Gaussian random process • Hence, the effective noise + interference for each user can be viewed as an additive white Gaussian noise! • Remark: the assumption that each interferer contributes a roughly equal small fraction to the total interference is valid due to tight power control in CDMA 21
Processing ¡Gain • Received energy per chip: h [ m ] | 2 i | h ( k ) | x k [ m ] | 2 ⇤ X E c ⇥ k := E E l l • SINR per chip: small E c 1 SINR 1 ,c := k 6 =1 E c P k + σ 2 • SINR per bit: E b 1 || u || 2 E c G E c 1 1 SINR 1 ,b := k + σ 2 = k 6 =1 E c k 6 =1 E c P P k + σ 2 ⇤ T ⇥ s I s I s I 1 [0] 1 [1] 1 [ G − 1] u = · · · • G : Processing Gain 22
IS-‑95 ¡Uplink ¡Architecture Processing gain PN Code = 1238.8/9.6 = 128 Generator for I channel 1.2288 Mchips/s Baseband Forward Link Shaping Data Filter –90 ˚ 9.6 kbps Rate = 1/3, K = 9 64-ary Output 4.8 kbps Block Repetition Carrier 1.2288 Mchips/s Convolutional Orthogonal CDMA 2.4 kbps Interleaver × 4 Generator Encoder Modulator Signal 1.2 kbps 28.8 Baseband ksym / s Shaping Filter 1.2288 Mchips/s PN Code Generator for Q channel 23
Power ¡Control • Maintain equal received power for all users in the cell • Tough problem since the dynamic range is very wide. Users’ attenuation can differ by many 10’s of dB • Consists of both open-loop and closed loop - Open loop sets a reference point - Closed loop is needed since IS-95 is FDD • Consists of 1-bit up-down feedback at 800 Hz • Consumes about 10% of capacity in IS-95 • Latency in access due to slow powering up of mobiles 24
Power ¡Control ¡Architecture Initial downlink Received Transmitted power power signal Estimate measurement Channel uplink power Measured required SINR Outer loop Inner loop Open loop ±1dB Measured Measured Update error probability Frame SINR < or > β β > or < target decoder rate Closed loop 25
Interferene ¡Averaging • The received SINR for a user: P SINR = N 0 + ( K − 1) P + P ∈ cell I i i/ • In a large system, each interferer contributes a small fraction of the total out-of-cell interference - Made possible due to power control • This can be viewed as providing interference diversity • Same interference-averaging principle applies to voice bursty activity and imperfect power control 26
Soft ¡Handoff • Provides another form of diversity: macrodiversity - Two base stations can simultaneously decode the data Switching center Power control bits ± 1 dB ± 1 dB Base-station 1 Base-station 2 Mobile 27
Recommend
More recommend
