Scheduling and capacity estimation in LTE Olav Østerbø, Telenor CD (Corporate Development)
Agenda Introduction Obtainable bitrate as function of SINR Radio channel propagation model Radio signal fading model Analytical models for LTE radio network performance Numerical examples Conclusions
Introduction LTE is the new, (following up of HSPA) mobile access technology specified by 3GPP: Flat all-ip architecture Flexible in frequency bands (700 MHz-.6 GHz) Flexible carrier bandwidths (.4, 3, 5,, 5 and 0 MHz) Increased spectrum efficiency based on OFDMA for uplink, SC-FDMA for downlink Typical cell capacity (0 MHz bandwidth), 0 40 Mbps for downlink link and 5 5 Mbps for uplink Momentum in the industry, building on current investments in the GSM/UMTS 3
6D Obtainable bitrate as function of SINR Bitrate function B=B(SINR) Upper bound Shannon: B/f=Log (+SINR) Discrete; B/f based on CQI-table (3GPP TS 36.3) and Linear relation between SINR[dB] and CQI-index Approximate/Truncated modified version of Shannon's formula: B/f=MIN[T, C Log (+γsinr)] CQI index modulation code rate x 4 efficiency 0 out of range QPSK 78 0.53 QPSK 0.344 3 QPSK 93 0.3770 4 QPSK 308 0.606 5 QPSK 449 0.8770 6 QPSK 60.758 7 6QAM 378.4766 8 6QAM 490.94 9 6QAM 66.4063 64QAM 466.7305 64QAM 567 3.33 64QAM 666 3.903 3 64QAM 77 4.534 4 64QAM 873 5.5 5 64QAM 948 5.5547 H D @t b i s z N ormalised T hroughpu t 5 4 3 ---- Shannon ---- Modified Shannon ---- LTE CQI table 0 5 5 0 5 SINR@dBD 4
Radio channel propagation model r Signal-to-noise ratio: P w sending power Path loss (model): SINR = G = L P G w C and A constants, X t shadowing usually assumed to be normal distributed with zero mean and given standard deviation Noise N = N int + N ext sum the internal (or own-cell) noise power and is the external (or other-cell) interference. N L = C Alog ( r) + X t 5
Radio signal fading model SINR on the form: Stochastic part of SINR: r α h(λ) Slow fading (shadowing): Lognormal S t t ln X e Fast fading: Rayleigh, i.e. neg. exp. distributed St : Suzuki distributed with CDF S = X X ln X e (ln( t x)) e S ~ t xt su ( x) = e sln ( t) dt = σ t t= 0 πσ t= 0 dt Truncated version: shxl 3.5.5 σ =0. Lognormal (ln( t x)) T t ~ σ k k σ e ( ) k k σ Ssu ( x, T ) = dt = x e erfc( + πσ t 0 k! t= 0 k= σ σ =5.0 σ =.0 shxl.75.5.5 Suzuki σ =5.0 σ =.0 σ =.0 ln( x T ) ) 0.5 σ =.0 σ =0.6 0.75 0.5 0.5 σ =0.6 σ =0. 6 3 4 5 x 3 4 5 x
Analytical models for LTE radio network performance Spectrum efficiency through the bit-rate distribution per Recourse Block (RB) for users that are either randomly or located at a particular distance in a cell. Cell throughput/capacity and fairness by taking the scheduling into account. Scheduling based on metrics which depends (only) on own SINR and distance Specific models for the common (basic) scheduling algorithms, Round Robin, Proportional Fair and Max-SINR. Estimation of the capacity usage for GBR sources in LTE Non-persistent allocation, i.e. allocation every TTI to obtain GBR rate Cell throughput/capacity for a mix of GBR and Non-GBR (greedy) users 7
Input parameters to numerical examples Parameters Bandwidth per Resource Block Total Numbers of Resource Blocks -dependent path loss. (Taken from a 3GPP document) Numerical values 80 khz=x 5 khz 0 RBs L=C +37.6log (r), r in kilometers and C=8. db for GHz Lognormal Shadowing with standard deviation 8 db (in most of the cases) Rayleigh fast fading Noise power at the receiver - dbm Total send power 46.0 dbm=(40w) Radio signaling overhead 3/4 8
Mean throughput per RB as function of cell radius sd @t M bi p ut p e r R B M ean through p 0.8 0.7 0.6 0.5 0.4 0.3 0. Located at cell edge Suzuki distributed fading, GHz frequency and σ=0db, db, 5dB, 8dB, db from below. Random location 0. 3 4 5 Throughput per RB drops for large cells. Approx 0. Mbit/s for km cell and 0.05 Mbit/s at cell edge with 8dB shadowing. 9
Multiuser gain as function of cell radius M sd @t bi acity C ell ca p 80 70 60 50 40 30 0 GHz frequency, 0 RBs, fading Susuki distributed with shadowing σ=8 db, number of users =,, 3, 5,, 5, 0 from below. -- PF 3 4 5 -- Max-SINR -- RR Multiuser gain very large for Max-SINR. PF doubles cell throughput compare to RR for cell of km and 5 users.
Mean Bitrate for a user located at cell edge as function of cell radius. M sd @t bi C apacity p e r u se r 40 35 30 5 0 5 5 Max- SINR PF RR -users 3 4 5 M sd @t bi C apacity p e r u se r 5 0 5 5 Max- SINR PF 3-users RR 3 4 5 M sd @t bi C apacity p e r u se r 4 D 6 8 6 4 Max- SINR PF 5-users RR 3 4 5 M sd @t bi C apacity p e r u se r 7 D 8 6 5 4 3 Max- SINR PF -users RR 3 4 5 Max-SINR shows very poor cell edge performance Scheduling: RR, PF and Max-SINR scheduling algorithm, GHz frequency and 0 RBs
Mean cell throughput for users scheduled according to PF and a GBR user 80 80 sd @t M bi C ell c apacit y 70 60 50 40 30 0 -- Non-Persistent, cell edge -- Non-Persistent, random -- mean PF users GBR=3 Mbit/s sd @t M bi C ell c apacit y 70 60 50 40 30 0 -- Non-Persistent, cell edge -- Non-Persistent, random -- mean PF users GBR= Mbit/s 0.5.5.5 0.5.5.5 80 80 sd @t M bi 70 60 50 -- Non-Persistent, cell edge 70 -- Non-Persistent, cell edge -- Non-Persistent, random -- Non-Persistent, random -- mean PF users 60 -- mean PF users sd @t M bi 50 C ell c apacit y 40 30 0 GBR=0.3 Mbit/s C ell c apacit y 40 30 0 GBR=0. Mbit/s 0.5.5.5 GBR rates of less than Mbit/s does not reduce the overall throughput very much. GBR rates larger than Mbit/s is not recommended. 0.5.5.5 GBR of 3.0,.0, 0.3, 0. Mbit/s using non-persistent scheduling, for GHz and 0 RB and Suzuki distributed fading with std. σ=8db.
Conclusions The two most important factors for the radio performance in LTE are fading and attenuation due to distance. Numerical examples for LTE downlink shows results which are reasonable; In the range 5-50 Mbit/s for km cell radius at GHz with 0 RBs. Multiuser gain is large for the Max-SINR algorithm but also the PF algorithm gives relative large gain relative to plain RR. The Max-SINR has the weakness that it is highly unfair in its behaviour. (Not recommended to use in real operation.) The usage of GBR with high rates may cause problems in LTE due to the high demand for radio resources if users have low SINR i.e. at cell edge. GBR rate limited to at most Mbit/s per user? 3