基於行動電信網路的車速量測之研究

(1)

國

立

交

通

大

學

網路工程研究所

碩

士

論

文

基於行動電信網路的車速量測之研究

Deriving the Vehicle Speeds from Mobile

Telecommunications Network

研究生：張裕隆

指導教授：林一平教授

(2)

基於行動電信網路的車速量測之研究

Deriving the Vehicle Speeds from Mobile Telecommunication Network

研究生：張裕隆 Student：Yu-Long Chang

指導教授：林一平 Advisor：Yi-Bing Lin

國立交通大學

網路工程研究所

碩士論文

A Thesis

Submitted to Institute of Network Engineering

College of Computer Science

National Chiao Tung University

in partial Fulfillment of the Requirements

for the Degree of

Master

in

Computer Science

June 2013

Hsinchu, Taiwan, Republic of China

(3)

基於行動電信網路的車速量測之研究

網路

研究

(Intelligent Transportation Systems ITS) 測 (sensors)

測量路車速 Lin-Chang-Huangfu (LCH)

行動 (Mobile Switching Center MSC) 的

電信網路 LCH 的

(cell residence times) 的 (bias) 的

LCH 於車速的的車

測 (Vehicle Detector VD) 的的研究 LCH

車速的的

(4)

Deriving the Vehicle Speeds from Mobile

Telecommunications Network

Student: Yu-Long Chang

Advisor: Dr. Yi-Bing Lin

Institute of Network Engineering

National Chiao Tung University

ABSTRACT

Vehicle speeds of roads are often measured by the Intelligent Transportation Systems (ITS) through some sensors or software solutions. Our previous work proposed the Lin-Chang-Huangfu (LCH) scheme to compute the cell residence times by the standard counter values in the mobile telecommunications switches without any modiﬁcation to the telecommunications network. In this paper, we investigate the accuracy of the LCH scheme by deriving the bias of the cell res-idence times computed in this scheme. Then we extend the LCH scheme with some ﬁltering techniques for the vehicle speed estimation, and validate our approach with the vehicle detector measurements at National Highway 3, Longtan Township, Taoyuan County, Taiwan. Our study indicates that the LCH scheme is an effective approach for the vehicle speed estimation.

Index Terms: Lin-Chang-Huangfu (LCH) Scheme, Mobile Switching Center (MSC), telecom-munication, vehicle speed

(5)

誌

謝

謝的的的研究的的謝謝的的謝的謝的謝謝的的謝的謝謝的的的謝的的路謝謝 Hank 謝的謝謝的謝的謝謝的謝謝

(6)

· · · · i ABSTRACT · · · · ii 誌謝 · · · iii · · · · iv · · · · v · · · · vi · · · · 1 Lin-Chang-Huangfu (LCH) · · · · 4 的 (Bias) · · · · 6 車速之 _{· · · 16} · · · 20 · · · 25 · · · 26 A 的 · · · 29 B E[ρn(τ )]之 · · · 31 C · · · 33

(7)

1.1 . . . 1 1.2 電信網路 . . . 2 3.1 動的 (△: 動 i N: i ◦: 電 _•: 電 : ∆τ ) . . . . 7 4.1 車速的 . . . 17 5.1 ( × VD ◦ BS • 的 ) . . . 20 5.2 2011 9 16 VD LCH 的車速 (∆t = 1 hour x = 1 km) . 22 5.3 2011 9 25 VD LCH 的車速 (∆t = 1 hour x = 1 km) . 23 5.4 LCH VD 之的 (∆t = 1 hour x = 1 km) . . . . 24 A.1 的 . . . 29

(8)

∆t ∆τ τ 的 (⌊ τ ∆t ⌋ ∆t, ⌊ _τ ∆t+ 1 ⌋ ∆t ) ρ(τ ) ∆τ 的量

tc (call holding time)

tm (cell residence time)

t∗_m(τ ) LCH ∆τ 的的

N (τ ) ∆τ 的

λu(τ ) ∆τ 的速

ta (inter-call arrival time)

v(τ ) 路 ∆τ 的車速 btm(t∗m(τ )) 的 α(τ ) ∆τ 的 β(τ ) ∆τ 的的 γ(τ ) ∆τ 的 λ 速 1 η 的 1 µ 的

(9)

1 δ ∆τ 的 fm(tm) 的 (density function) Fm(tm) 的 (distribution function) f_m∗(tm) fm(tm)的 τc τm Pi i 的 q1 tc> τm的的 q2 τc> tm的的的 ρn(τ ) ∆τ 的 ρe(τ ) ∆τ 的 ρh(τ ) ∆τ 的 ∆τn (new call) 的 ∆τ 的 ∆τh (handover call) 的 ∆τ 的 t t∗ t 的 n1 電的 n2 tc,i i 的 fc(tc,i) i 的的 θ t n2 量測的 ∆τk k 的

(10)

Kh 車速的

Vs (near-history) 的

(11)

的 (Intelligent Transportation Systems ITS) 測量路車速

行車車路 ITS

路車速網路 ( Internet) 1.1

的

(12)

Cell i-1

Cell i-2

Cell i

Cell i+1

Cell i+2

Road

Internet

Mobile Core Network

Network

Probe

ITS Server

b

c

d

f

g

h

a

MSC

e

1.2: 電信網路測量車速的 : • 車測 (Vehicle Detector VD) [1] 車測 ( 1.2⃝)a 路

(13)

測量車 ( 1.2⃝) 的速b 網路車速 ITS ( 1.2⃝)c

• 基於的測車 (Global Positioning System(GPS)-based Vehicle Probe GVP) [2, 3] 車的 (User Equipments UEs) GPS

GPS 行動電信網路基

(Base Stations BSs 1.2⃝)d 行動 (Mobile Switching Center MSC

1.2⃝)e ITS 的 GPS 車速

• 動車的信 (Cellular Floating Vehicle Data CFVD) [4, 5, 6] 網路測 (Network Probe 1.2 ⃝)f BSs MSC 之的信

(signals) 信 ITS 的行 ITS

的 (cell BS 的電 ) 的

(cell residence times) 動速

VD 的測的的 [4, 6] GVP GPS 的的電 GPS 的路 GPS 的車路的 GPS CFVD 的電信網路的信路 (signaling links) 網路測測行動網路之的信 (signaling messages) 的 [7, 8, 9] Lin-Chang-Huangfu (LCH) CFVD CFVD 的 LCH MSC 的 ( 的 ) 的 MSC 動的的速 LCH 的車速的 VD LCH LCH 的的的研究 LCH 車速的的 LCH LCH 的的車速的的 LCH 的之

(14)

Lin-Chang-Huangfu (LCH)

[7] LCH 1.2 的 MSC ( 1.2⃝)e 行動 [11] BS MSC BS 的電 ( 的 1.2⃝)g 的 BS MSC 動的路的的 (handover) 的行動電信 MSC 的行 ( / 電動 ) MSC ∆t 行的 15 的的 ( Erlang) τ ∆τ (⌊ _τ ∆t ⌋ ∆t, ⌊ _τ ∆t + 1 ⌋ ∆t ) 車速的路路 (target road segment 1.2 ⃝)h

i ( 1.2⃝) 的g ρ(τ ) ∆τ i的量

(carried traffic) ρ(τ ) ∆τ i的的 (carried call holding times )

(offered call holding time) 的 BS 的電

電的 ( ) 的 BS 的量的的電 MSC ρ(τ ) tc MSC 量測的 tc 的 ( tc ) γ(τ ) ∆τ i 的 tm [7] Little's Law[12]

(15)

∆τ i 的 t∗_m(τ ) Little's Law

N (τ ) = λu(τ )t∗m(τ ) (2.1)

N (τ ) ∆τ i 的 λu(τ ) ∆τ

i的速 ta (inter-call arrival time)

∆τ (∆t/ta) tc(∆t/ta) ρ(τ ) = N (τ )tc∆t ta N (τ ) = ρ(τ )ta tc∆t (2.2) 於 λu(τ ) ∆τ i的速 ∆τ λu(τ )∆t i (tc/ta) ∆τ (λu(τ )tc∆t/ta) i γ(τ ) = λu(τ )tc∆t ta λu(τ ) = γ(τ )ta tc∆t (2.3) (2.1) (2.2) (2.3) t∗_m(τ ) = ρ(τ ) γ(τ ) (2.4) 基於 (2.4) 路的車速 i− 1 的車 ( 1.2⃝)b i i + 1 x i 的路 (2.4) 路 ∆τ 的 車速 v(τ ) v(τ ) = x t∗_m(τ ) = xγ(τ ) ρ(τ ) (2.5) LCH 的的 LCH 車速的量

(16)

的

(Bias)

LCH 的的 LCH 於車速的的 LCH 車速的 tm 的 (2.4) 的 t∗m(τ ) tm 的的 btm(t∗m(τ )) = E[t∗m(τ )]− E[tm] (3.1) 的 (error) 的的的之的 [13] ∆τ 的 γ(τ ) ρ(τ ) 的 γ(τ ) ρ(τ ) t∗_m(τ ) (3.1) 3.1 6 [t0, t19] 動的 3.1 UE 3 的行 UE 3 t8 動 i ( △ ) t11 電 UE 3 ( ◦ ) UE 3 t13 i ( N ) t15 UE 3 的 ( • ) UE 3 的 tm = t13− t8 UE 3 的 tc= t15− t11 ∆τ = t19− t5 ( 3.1 ) ∆τ i 測的的 : • (new call) 於 ∆τ 的 3.1 的 t10 UE 6 的 t11 UE 3 的 • 的 (existing call) ∆τ i的 UE 1 的電 ∆τ 的 t4 UE 1 UE 1

(17)

i UE 1 t5 i t4 < t5 行 • (handover call) 的 UE 2 t6 UE 4 t9 UE 5 t12 v v v

time

v v v 3.1: 動的 (△: 動 i N: i ◦: 電 •: 電 : ∆τ ) α(τ ) β(τ ) γ(τ ) 的 ∆τ i 的 3.1 α(τ ) = 2 ( t10 t11 的電 ) β(τ ) = 1 ( t4

(18)

的電 ) γ(τ ) = 3 ( t6 t9 t12的 ) λ 速 (call arrival rate) E[α(τ )] = λ∆τ (3.2) tm (mean) 1 η 的 tc 1 µ 的 (facts)

Fact 1. tc (exponential distribution) tm

(arbitrary density function) fm(tm) 速 η ∆τ

(arbitrary distribution) 1 δ E[β(τ )] = λ µ E[γ(τ )] = λη δµ Proof: ∆τ 的 i 的行的 E[β(τ )] E[α(τ )] 1 µ E[∆τ ] E[β(τ )] E[β(τ )] = E[α(τ )](1/µ) E[∆τ ] = λ µ (3.3) 1 µ < E[∆τ ] 1 µE[∆τ ] ∆τ 的於的 量 E[β(τ )] E[α(τ )] 1 µE[∆τ ] (3.3) E[γ(τ )] E[α(τ )] 電的於

的 (memoryless) 的 (residual call holding time 3.1 的 τc= t15−t13)

1

µ tm的

(distribution function) Fm(tm) (residual cell residence

time) τm( 3.1 的 τm = t13−t11) (residual life theorem)

[14] τm 的 (density function) 1− Fm(tm) (1/η) = [1− Fm(tm)]η 的量測 的 tc 的的電 Pi i 的 q1 = Pr[tc > τm] 的 q2 = Pr[τc > tm] 的的 Pi = q1q2i−1(1− q2) i≥ 1 於 電的 ∞ ∑ i=1 iPi = ∞ ∑ i=1 iq1q2i−1(1− q2) = q1 1− q2 (3.4)

(19)

(3.4) q1的 q1 = Pr[tc > τm] = ∫ _∞ τm=0 ∫ _∞ tc=τm µe−µtc_[1− F m(τm)]ηdtcdτm = ∫ _∞ τm=0 e−µτm_[1− F m(τm)]ηdτm (3.5) 1− q2 1− q2 = Pr[τc< tm] = ∫ _∞ τc=0 ∫ _∞ tm=τc µe−µτc_f m(tm)dtmdτc = ∫ _∞ τc=0 µe−µτc_[1− F m(τc)]dτc = ( µ η ) q1 (3.6) (3.5) (3.6) 電的 q1 1− q2 = η µ E[γ(τ )] E[γ(τ )] = E[α(τ )] ( η µ ) = λη δµ (3.7) 3.2 3.3 3.7 A的的

E[α(τ )] E[β(τ )] E[γ(τ )]的 ∆τ 的的 Fact 2. tc ∆τ 1 µ 1 δ tm fm(tm) 速 η 於 E[ρ(τ )] = λ δµ Proof: Fact 1 的 τc tc的 τm tm 的於 i的 ∆τ ρ(τ ) 的 3.1 ∆τ UE 3 的 t11 tc= t15− t11 t11 UE 3 的 τm = t13− t11 於的 (Poisson Process) t11 UE 3 的 (random observer) t11 ∆τ 的的 t11 的 ∆τ ∆τn 1 δ ∆τ 的 E[ρn(τ )]

(20)

E[ρn(τ )] = E[min(tc, τm, ∆τn)] = E[min(min(tc, ∆τn), τm)] (3.8) xn = min(tc, ∆τn) 速 µ + δ fm(tm)的 f_m∗(tm) B 的 (3.8) E[ρn(τ )] = E[min(xn, τm)] = 1 µ + δ − [ η (µ + δ)2 ] [1− f_m∗(µ + δ)] (3.9) 3.1 ∆τ UE 1 的 t4 t5 τc= t16− t5 τm = t18− t5 於 t5 UE 1 的 tc的 τc 1 µ 於 ∆τ 的 E[ρe(τ )] = E[min(τc, τm, ∆τ )] 於 τc ∆τ 速 µ δ E[ρe(τ )] (3.9) E[ρe(τ )] = E[ρn(τ )] 3.1 ∆τ UE 4 的 t9 i tm = t17− t9 t9 的 τc = t14− t9 於 t9 UE 4 tc的 tc 1 µ τc tc 的 t9 ∆τ 的 t9 ∆τ 的 ∆τh 1 δ ∆τ 的 E[ρh(τ )] E[ρh(τ )] = E[min(τc, tm, ∆τh)] = E[min(min(τc, ∆τh), tm)] (3.10) xh xh = min(τc, ∆τh) 速 µ + δ 於

(21)

(3.9) 的 (3.10) E[ρh(τ )] = E[min(xh, tm)] = ( 1 µ + δ ) [1− f_m∗(µ + δ)] (3.11) (3.2) (3.9) (3.11) Fact 1 E[ρ(τ )]

E[ρ(τ )] = E[α(τ )]E[ρn(τ )] + E[β(τ )]E[ρe(τ )] + E[γ(τ )]E[ρh(τ )]

= (E[α(τ )] + E[β(τ )]) { 1 µ + δ − [ η (µ + δ)2 ] [1− f_m∗(µ + δ)] } +E[γ(τ )] {( 1 µ + δ ) [1− f_m∗(µ + δ)] } = λ δµ Fact 3. ∆τ 1 δ tc tm E[ρ(τ )] = λ δµ Proof: Fact 2的 ρ(τ ) ∆τ 的 E[ρn(τ )] E[ρn(τ )] = E[min(yn, ∆τn)] = E[min(min(tc, τm), ∆τn)] (3.12) (3.8) ∆τn [ 0,1 δ ] 的 (uniform distribution) δ 於 tc τm 速 µ η yn = min(tc, τm) 速 µ + η (3.12) E[ρn(τ )] = ∫ 1 δ ∆τn=0 ∫ ∆τn yn=0

δyn(µ + η)e−(µ+η)yndynd∆τn

+ ∫ 1 δ ∆τn=0 ∫ _∞ yn=∆τn δ∆τn(µ + η)e−(µ+η)yndynd∆τn = ( 1 µ + η ) [ 1− δ µ + η + δe−(µ+ηδ ) µ + η ] (3.13)

(22)

∆τ 的 E[ρh(τ )] E[ρh(τ )] = E[min(τc, tm, ∆τh)] (3.14) τc tm 速 µ η ∆τh [ 0,1 δ ] 的 (3.14) 的 (3.12) E[ρh(τ )] = E[ρn(τ )] ∆τ 的 E[ρe(τ )] E[ρe(τ )] = E [ min ( ye, 1 δ )] = E [ min ( min(τc, τm), 1 δ )] (3.15) ye = min(τc, τm) 速 µ + η (3.15) E[ρe(τ )] = ∫ 1 δ ye=0 ye(µ + η)e−(µ+η)yedye+ ∫ _∞ ye=1_δ ( 1 δ ) (µ + η)e−(µ+η)ye_dy e = ( 1 µ + η ) [1− e−(µ+ηδ )] _(3.16) (3.2) (3.13) (3.16) Fact 1 E[ρ(τ )]

E[ρ(τ )] = E[α(τ )]E[ρn(τ )] + E[β(τ )]E[ρe(τ )] + E[γ(τ )]E[ρh(τ )]

= (E[α(τ )] + E[γ(τ )]) ( 1 µ + η ) [ 1− δ µ + η + δe−(µ+ηδ ) µ + η ] +E[β(τ )] ( 1 µ + η ) [ 1− e−(µ+ηδ ) ] = λ δµ Fact 4. ∆τ tc tm 1 δ 1 µ 1 η 於 E[γ(τ )]≈ λη δµ Proof: 的 t ≫ 1 δ 1 µ t ∗ _t 的於 t ⌈ t∗ (1/µ) ⌉ =⌈µt∗⌉ 的 t ⌊ t (1/η) ⌋ ⌊ t∗ (1/η) ⌋ 的 ⌊ t∗ (1/η) ⌋ =⌊ηt∗⌋

(23)

電的 E[n1] = ⌊ηt ∗_⌋

⌈µt∗_⌉ limt∗→∞E[n1] =

η

µ t

E[γ(τ )] = E[α(τ )]E[n1]≈

λη δµ (3.17) Fact 5. ∆τ = 1 δ tc tm 1 µ 1 η t 於 limt→∞E[ρ(τ )] = λ δµ Proof: n2 動 t≫ n2 δ , n2 µ 的行於 n2 的 tm的的 t ⌈λtn2⌉ n2 t ( initial effect) 的 t

( end effect) tc,i i 的

fc(tc,i) i 的的速 µ 於 t n2 量測的 θ θ = ⌈λtn∑2⌉ i=1 ∫ _∞ tc,i=0

tc,ifc(tc,i)dtc,i = ⌈λtn 2⌉ µ n2 ⌈ t (1/δ) ⌉ =⌈n2tδ⌉ 於 E[ρ(τ )] ∆τ i的 E[ρ(τ )]≈ θ ⌈n2tδ⌉ t lim t→∞E[ρ(τ )] = λ δµ (3.18)

的 Fact 5的 Fact 2 Fact 3 (Erlang

equation) 的量 λ µ ∆τ 的量 λ δµ Fact 6. tc tm 1 µ 1 η ∆τ = 1 δ t 的行於 btm(t∗m(τ )) = µδ λη2 Proof: t→ ∞ 的行於 LCH 的量測 ∆τ

(24)

行 (3.17) (3.18) 1 δ 於 E[γ(τ )] E[ρ(τ )] (2.4) t∗_m(τ ) c1 = E[ρ(τ )] (1/δ) = λ µ c2 = E[γ(τ )] (1/δ) = λη µ 於 (2.4) f (r, s) t∗_m(τ ) = f (r, s)|r=c1,s=c2 = (_r s ) r=c1,s=c2 (3.19) (3.19) (Taylor series) f (r, s) (c1, c2)的 f (r, s) = f (c1, c2) + [ ∂f (r, s) ∂r r=c1,s=c2 ] (r− c1) + [ ∂f (r, s) ∂s r=c1,s=c2 ] (s− c2) + ( 1 2 ) [ ∂2_{f (r, s)} ∂r2 r=c1,s=c2 ] (r− c1)2 + [ ∂ ∂s ( ∂f (r, s) ∂r ) r=c1,s=c2 ] (r− c1)(s− c2) + ( 1 2 ) [ ∂2f (r, s) ∂s2 r=c1,s=c2 ] (s− c2)2+· · · (3.20) 於 f (r, s) (c1, c2) 的 E[r] = c1 E[s] = c2

E[r− c1] = 0, E[s− c2] = 0 E[(r− c1)(s− c2)] = 0 (3.21)

(3.19) ∂2_{f (r, s)} ∂r2 r=c1,s=c2 = 0 ∂ 2_{f (r, s)} ∂s2 r=c1,s=c2 = 2c1 c3 2 (3.22) (3.21) (3.22) (3.20) 的 E[f (r, s)] = f (c1, c2) + ( 1 2 ) [ ∂2_{f (r, s)} ∂s2 r=c1,s=c2 ] E[(s− c2)2] = c1 c2 + ( c1 c3 2 ) E[(s− c2)2] = 1 η + ( µ2 λ2_η3 ) E[(s− c2)2] (3.23)

(25)

(3.23) E[(s− c2)2] = V ar [ γ(τ ) (1/δ) ] = δ2V ar[γ(τ )] (3.7)

γ(τ ) (Poisson random variable) E[γ(τ )] = λη δµ (variance) V ar[γ(τ )] = λη δµ (3.23) E[f (r, s)] = 1 η + ( µ2 λ2_η3 ) V ar    (γ(τ )1 δ )     = 1 η + µδ λη2 (3.24) 於 tm 1 η (3.1) (3.24) btm(t∗m(τ )) btm(t ∗ m(τ )) = E[t∗m(τ )]− E[tm] = [ 1 η + µδ λη2 ] − 1 η = µδ λη2 (3.25) (3.25) LCH ( µ ) ∆τ ( δ ) 速 λ ( η ) 的車行的 ( 的速 ) (3.25) LCH 車速行的的

(26)

車速

之

(2.5) 車速的路的 (leaky-bucket) [16] 的 i ( 4.1(a) 的⃝)1 路之的 ( 4.1(a) 的⃝2 ⃝)3 路的的行 γ(τ ) ρ(τ ) 行 (2.5) 的行動電信網路的 (location update) i 路的

行動電信網路 (Location Areas LAs LA j

i− 1 i 4.1(a) 的⃝)4 ( 4.1(a) 的⃝)5 的

LA 動 LA 行 MSC 的 LA [11, 15]

BS MSC 行動 ( Visitor Location Register VLR)

基於路的 : 1. 於 i 行的 i 的 LA A 行的 i 的 LA B 行的 A B 路路動 LA j− 1 LA j 動 LA j + 1 ( 4.1(a) ⃝)5 LA j 的 i− 1 ( A) LA j + 1 的 i + 1 ( B) i− 1 i + 1 路路行的車的車 i 的 LA 路路 ( 4.1(b) ⃝)1 1 的測 :

(27)

g

R

oa

d 1

Cell

i

Cell

i+1

Cell

i+2

Cell

i-3

Cell

i-2

Cell

i-1

LA j-1

LA j

LA j+1

1

5

4 Road 2

2

3

(a) 路行

Cell

i

Cell

i+1

Cell

i+2

Cell

i-3

Cell

i-2

Cell

i-1

LA j-1

LA j

LA j+1

Cell

i+3

1

(b) 車動路的

(Location Area LA)

(28)

2. 於 i 行的的路的路動 4.1(b) 的 {i− 2, i − 1, i} {i− 1, i, i + 1} {i, i + 1, i + 2} 路 1 2 的 i的 路的 γ(τ ) ρ(τ ) 於路的速 (2.5) 的路的的量 ∆τ 的路的車速基於的 (far-history) (near-history) ∆τ 的的量的的的的於 ∆τ ∆τk k 的 ∆τ0 = ∆τ ∆τ 的 ∆τ1 的 Kh v(τ ) . Kh 的路速 v(τ ) = x[∑K_k=0γ(τk) ] ∑K k=0ρ(τk) where K = min { N : N ∑ n=0 γ(τn)≥ Kh } (4.1) 的 γ(τ ) 於 Kh (4.1) 的 (2.5) ( ) MSC ∆t ρ(τ ) γ(τ ) ( ) 的的的的車速的 ( 的車速於 Vs)

(29)

. 路速 v(τ )←    W v(τ ) + (1− W )v(τ − ∆t), for |v(τ) − v(τ − ∆t)| < Vs v(τ ), for|v(τ) − v(τ − ∆t)| ≥ Vs (4.2) 0≤ W ≤ 1 (weighting factor) v(τ − ∆t) ∆τ 的車速 Vs 測的 (4.2) 的 W 的車速 v(τ − ∆t) 於 的車速 v(τ ) 的

(30)

LCH 的車速 VD 量測的車速 ( 5.1 的⃝) 的車速a 車速速路 66 km 的 VD ( 5.1 的⃝) 量測b 的 BS ( 速路 66.8 km VD 800 5.1 的⃝)c LCH 1 2 車速的 x = 1 km ∆t = 1 hour x的 [17, 18] 測量 Kh = 10 Vs = 40 km/hour W = 0.5

VD 66 km

Base Station 66.8 km

a

b

c

5.1: ( × VD ◦ BS • 的 )

(31)

vs VD 量測的車速 (s = V D) LCH 的車速 (s = LCH) 5.2(a) 5.2(b) 路 2011 9 16 8:00 20:00 的 vs VD LCH 8:00 20:00 的 ( vs > 75 km/hour) 5.3(a) 5.3(b) 路 2011 9 25 8:00 20:00 的 vs 5.3(a) 測 18:00 20:00 的 ( vs< 60 km/hour) 5.3(b) 8:00 20:00 的 ( vs > 75 km/hour) 5.2 5.3 車行 LCH 的速 VD (discrepancy) ϵ ϵ =vLCH − vV D vV D (5.1) 基於 5.2 5.3 5.4 的 ϵ ϵ 的於 20% ( 5.4(b) 19:00 20:00 的 ◦ ) ϵ 於 20% 車的 (3.25) LCH 的的的行動電信網路 49 (2011 9 13 2011 10 31 ) 8:00 20:00 的 γ(τ ) ρ(τ ) 基於的研究的 E[ϵ] 14.46% 1 2 E[ϵ] 12.7% E[ϵ] 7.51%

(32)

0 25 50 75 100 125 vs (km/hour) 8 10 12 14 16 18 20 Time (hour) ◦: s = LCH ×: s = V D . ... ◦ ◦ _◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ _◦ ... × _× × × × × _× _× _× × _× × (a) 0 25 50 75 100 125 vs (km/hour) 8 10 12 14 16 18 20 Time (hour) ◦: s = LCH ×: s = V D . ... ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ _◦ ... × × × × × × × × × × × × (b) 5.2: 2011 9 16 VD LCH 的車速 (∆t = 1 hour x = 1 km)

(33)

0 25 50 75 100 125 vs (km/hour) 8 10 12 14 16 18 20 Time (hour) ◦: s = LCH ×: s = V D . . ... ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ _◦ ◦ ◦ ◦ . ... × × _× _× × _× × _× × × × × (a) 0 25 50 75 100 125 vs (km/hour) 8 10 12 14 16 18 20 Time (hour) ◦: s = LCH ×: s = V D . ... ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ . ... × × _× × × × × × × × _× _× (b) 5.3: 2011 9 25 VD LCH 的車速 (∆t = 1 hour x = 1 km)

(34)

0 20 40 60 80 100 ǫ(%) 8 10 12 14 16 18 20 Time (hour) ◦: Northbound •: Southbound . ... ◦ ◦ _◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ . . ... • • • • • • • • • • • • (a) 2011 9 16 的 0 20 40 60 80 100 ǫ(%) 8 10 12 14 16 18 20 Time (hour) ◦: Northbound •: Southbound . ... ◦ ◦ ◦ ◦ ◦ ◦ _◦ ◦ ◦ ◦ ◦ ◦ . ... • _• • • • • • • • • • • (b) 2011 9 25 的 5.4: LCH VD 之的 (∆t = 1 hour x = 1 km)

(35)

的 LCH [7, 8] 車速的 LCH 於車速的 ( 行的速 ) LCH 的的 LCH 之車速的 VD 量測之車速的 LCH 的車速 VD 量測的車速的的之的 (discrepancies) 於 20% 1 2 E[ϵ] 14.46% 12.7% E[ϵ] 7.51% 路的的車速的的研究 LCH 的測路的車速的 ( VD 的測 ) 車速研究 ITS 網路的車速測 [19] 的車速的

(36)

[1] Sang Jin Park, Tae Yong Kim, Sung Min Kang, and Kyung Heon Koo , "A novel sig-nal processing technique for vehicle detection radar," in 2003 IEEE MTT-S Internatiosig-nal

Microwave Symposium Digest, 2003, pp. 8-13.

[2] Anurak Poolsawat, Wasan Pattara-Atikom, and Boonchai Ngamwongwattana, "Acquiring road traffic information through mobile phones," in 8th International Conference on ITS

Telecommunications (ITST), 2008, pp. 170-174.

[3] Ruey Long Cheu, Chi Xie, and Der-Horng Lee, "Probe vehicle population and sample size for arterial speed estimation," Computer-Aided Civil and Infrastructure Engineering, vol. 17, no. 1, pp. 53–60, Jan. 2002.

[4] Chih-Yi Chiang, Ju-Yin Chuang, Jian-Kai Chen, Chia-Chen Hung, Wei-Hui Chen, and Kuen-Rong Lo, "Estimating instant traffic information by identifying handover patterns of UMTS signals," in 2011 14th International IEEE Conference on Intelligent Transportation

Systems (ITSC), 2011, pp. 390-395.

[5] Kuen-Rong Lo, Chih-Yi Chiang, Ju-Yin Chuang, Jian-Kai Chen, Chia-Chen Hung, and Wei-Hui Chen , "Feasibility analysis of UMTS handover logs for traffic state estimation," in 2011 11th International Conference on ITS Telecommunications (ITST), 2011, pp. 684-690.

[6] Bon-Yeh Lin, Chi-Hua Chen, and Chi-Chun Lo, "A traffic information estimation model using periodic location update events from cellular network", Communications in

Com-puter and Information Science, 2011, vol. 135, pp. 72-77.

[7] Yi-Bing Lin, Ming-Feng Chang, and Chien-Chun Huang-Fu , "Derivation of cell residence times from the counters of mobile telecommunications switches," IEEE Transactions on

(37)

Wireless Communications, 2011, vol. 10, no. 12, pp. 4048-4051.

[8] Yi-Bing Lin, Chien-Chun Huang-Fu, and Nabil Alrajeh, "Predicting human movement based on telecom's handoff in mobile networks," IEEE Transactions on Mobile Computing, in press.

[9] Chien-Chun Huang-Fu and Yi-Bing Lin, "Deriving Vehicle Speeds From Standard Statis-tics of Mobile Telecom Switches,"IEEE Transactions on Vehicular Technology, 2012, vol. 61, no. 7, pp. 3337-3341.

[10] Ren-Huang Liou, Yi-Bing Lin, Yu-Long Chang, and Ming-Feng Chang, "Deriving the ve-hicle speeds from mobile telecommunications network," the 12th International Conference

on ITS Telecommunications (ITST), November 2012.

[11] Yi-Bing Lin and Ai-Chun Pang, Wireless and Mobile All-IP Networks. John Wiley & Sons, Inc., 2005.

[12] Edward D. Lazowska, John Zahorjan, G. Scott Graham, and Kenneth C. Sevcik,

Quanti-tative System Performance Computer System Analysis Using Queueing Network Models.

Prentice-Hall, Inc., 1984.

[13] Jay L. Devore, Probability and Statistics for Engineering and the Sciences. Duxbury Press, 2011.

[14] Leonard Kleinrock, Queueing Systems, Vol. I, Theory, Wiley, New York, 1976.

[15] Yi-Bing Lin, "Performance modeling for mobile telephone networks", IEEE Network, vol. 11, pp. 63–68, Nov./Dec. 1997.

[16] Yi-Bing Lin and Imrich Chlamtac, Wireless and Mobile Network Architectures. John Wiley & Sons, Inc., 2000.

[17] David Gundlegard and Johan M. Karlsson, "Handover Location Accuracy for Travel Time Estimation in GSM and UMTS," IET Intelligent Transport Systems, 2009, vol. 3, no. 1, pp. 87-94.

[18] Hillel Bar-Gera, "Evaluation of a Cellular Phone-Based System for Measurements of Traf-fic Speeds and Travel Times: A Case Study from Israel," Transpiration. Research Part C:

(38)

Emerging Technologies, 2007, vol. 15, no. 6, pp. 380-391.

[19] Chai-Hien Gan and Yi-Bing Lin, "Push-to-talk service for intelligent transportation sys-tems," IEEE Transactions on Intelligent Transportation Systems, 2007, vol. 8, no. 3, pp. 391-399.

(39)

A

的

If(flag==true) ++ ; Generate the next UE movement event and insert it in the event list

Generate the first call arrival,

UE movement and events, and insert them in the event list.

++; If(flag==true) ++ ; Generate the next event and insert it in the event list flag=false;

Generate the next call

arrival event and insert it in the event list ++ ; flag=true; Generate the next call completion event and insert it in the event list ; ck=1; flag=false; Start

Process the next event e. ck=e.timestamp;

e.type=?

End

(40)

A.1 的 3.2 3.3 3.7

1 的 Nn Ne Nh N∆t

0 (timestamp) ck 誌 f lag 1 f alse 2

的動 (event

list) 的的 3

的的 4

的 (call arrival) 的行 5 動 (UE

movement) 的行 6 (call completion) 行 7

(∆t) 的 行 8 5 Nn 1 f lag true 的行 3 6 f lag true Nh 1 行動的行 3 7 f lag f alse 的行 3 8 N∆t 1 f lag true Ne 1 行行 9 9 N∆t 106 行 10 行 3 10 的 Nn Ne Nh N∆t α(τ ) β(τ ) γ(τ ) 的

(41)

B

E[ρ

_n

(τ )]

之

[15] ∆τ 的 E[ρn(τ )] τm rm(t) Rm(t) rm∗(s) [14] rm(t) tm的 Fm(t) rm(t) = η[1− Fm(t)] (B.1) (B.1) tm 的 fm∗(s) rm∗(s)的 r∗_m(s) = ∫ _∞ t=0 η[1− Fm(t)]e−stdt = (_η s ) [1− f_m∗(s)] (B.2) Fρn(τ )(t) Fxn(t) ρn(τ ) xn = min(tc, ∆tn) 的 於 ρn(τ ) = min(xn, τm) Fρn(τ )(t) = Fxn(t) + Rm(t)[1− Fxn(t)] (B.3) ρn(τ ) 的 E[ρn(τ )] = ∫ _∞ t=0 [1− Fρn(τ )(t)]dt (B.4) (B.3) (B.4) E[ρn(τ )] =∈ t∞t=0[1− Fxn(t)]dt− ∫ _∞ t=0 Rm(t)[1− Fxn(t)]dt (B.5)

(42)

xn 速 µ + δ (B.2) (B.5) E[ρn(τ )] = E[xn]− ∫ _∞ t=0 Rm(t)e−(µ+δ)tdt = 1 µ + δ − r_m∗(µ + δ) µ + δ = 1 µ + δ − [ η (µ + δ)2 ] [1− f_m∗(µ + δ)] (B.6)

(43)

C

1 2 的 1 ifstream in1( RNC321 .c_str ()) ; 2 ifstream in2( RNC311 .c_str ()) ; 3 ifstream in3( RNC353 .c_str ()) ; 4 string tmp ; 5 int IMSI ; 6 string Time ; 7

8 while (in1 >> IMSI >> Time) 9 { 10 string type = "" ; 11 string subType = "" ; 12 int cellId ; 13 int sessionId ; 14 15 in1 >> tmp ; 16 type += tmp ; 17 while (in1 >> tmp) 18 { 19 if(tmp [0] >=65 && tmp [0] <=90) 20 break ; 21 type += " " ; 22 type += tmp ; 23 } 24 subType += tmp ; 25 while (in1 >> tmp )

(44)

26 { 27 if(tmp [0] >=48 && tmp [0] <=57) 28 break ; 29 subType += " " ; 30 subType += tmp ; 31 }

32 in1 >> cellId >> sessionId >> tmp >> tmp >> tmp >> tmp; 33 hour = atoi(Time. substr (0 ,2).c_str ()) ;

34 minute = atoi(Time. substr (3 ,2).c_str ()) ; 35 second = atoi(Time. substr (6 ,2).c_str ()) ; 36 int nowTime = hour *3600+ minute *60+ second ; 37

38 if( subType == " Normal loc updating ") 39 {

40 if( cellId %10000 == 3483 || cellId %10000 == 3488 || cellId

%10000 == 9929) // North

41 {

42 if(Nhead == NULL) 43 {

44 Nhead = new treenode (IMSI , nowTime ) ; 45 } 46 else 47 { 48 treenode *p = Nhead ; 49 while(p != NULL) 50 { 51 if(p->IMSI > IMSI) 52 { 53 if(p->left == NULL) 54 {

55 p->left = new treenode (IMSI , nowTime ) ; 56 break ;

57 }

(45)

59 {

60 p = p->left ;

61 }

62 }

63 else if(p->IMSI < IMSI)

64 {

65 if(p->right == NULL)

66 {

67 p->right = new treenode (IMSI , nowTime ) ; 68 break ; 69 } 70 else 71 { 72 p = p->right ; 73 } 74 } 75 else 76 { 77 break ; 78 } 79 } 80 } 81 }

82 else if( cellId %10000 == 3417 || cellId %10000 == 8494 ||

cellId %10000 == 3286) // South

83 {

84 if(Shead == NULL) 85 {

86 Shead = new treenode (IMSI , nowTime ) ; 87 }

88 else 89 {

90 treenode *p = Shead ; 91 while(p != NULL)

(46)

92 {

93 if(p->IMSI > IMSI)

94 {

95 if(p->left == NULL)

96 {

97 p->left = new treenode (IMSI , nowTime ) ; 98 break ; 99 } 100 else 101 { 102 p = p->left ; 103 } 104 }

106 {

108 {

109 p->right = new treenode (IMSI , nowTime ) ; 110 break ; 111 } 112 else 113 { 114 p = p->right ; 115 } 116 } 117 else 118 { 119 break ; 120 } 121 } 122 } 123 } 124 }

(47)

126 {

127 int x = -1 ;

128 if( cellId %10000 == pattern1 ) 129 {

130 x = 0 ; 131 }

132 else if( cellId %10000 == pattern2 ) 133 {

134 x = 1 ; 135 }

138 x = 2 ; 139 }

142 x = 3 ; 143 }

146 x = 4 ; 147 }

148 if(x >=0) 149 {

150 if( extrahead == NULL) 151 {

152 extrahead = new treenode (IMSI , nowTime ) ; 153 extrahead -> sessionId = sessionId ;

154 extrahead ->check[x] = 1 ; 155 extrahead ->count ++ ; 156 extrahead ->cell = x ; 157 } 158 else 159 {

(48)

160 treenode *p = extrahead ; 161 while(p != NULL) 162 { 163 if(p->IMSI > IMSI) 164 { 165 if(p->left == NULL) 166 {

167 p->left = new treenode (IMSI , nowTime ) ; 168 p->left -> sessionId = sessionId ;

169 p->left ->check[x] = 1 ; 170 p->left ->count ++ ; 171 p->left ->cell = x ; 172 break ; 173 } 174 else 175 { 176 p = p->left ; 177 } 178 }

180 {

182 {

183 p->right = new treenode (IMSI , nowTime ) ; 184 p->right -> sessionId = sessionId ;

185 p->right ->check[x] = 1 ; 186 p->right ->count ++ ; 187 p->right ->cell = x ; 188 break ; 189 } 190 else 191 { 192 p = p->right ; 193 }

(49)

194 } 195 else

196 {

197 p->count ++ ;

198 if(p-> sessionId == sessionId )

199 {

200 p->check[x] = 1 ;

201 if(x-p->cell == 0 && p->count <=3)

202 { 203 p->count -- ; 204 } 205 else if(p->count == 2) 206 { 207 if(x-p->cell >0) 208 { 209 p->dir =2 ; 210 } 211 else if(x-p->cell <0) 212 { 213 p->dir =1 ; 214 } 215 p->cell = x ; 216 } 217 else if(p->count == 3) 218 {

219 if((p->dir == 1 && x>p->cell) || (p->

dir == 2 && x<p->cell))

220 { 221 p->check[x] = 0 ; 222 p->count -- ; 223 } 224 } 225 } 226 else

(50)

227 {

228 for(int i=0;i< patternNum ;i++) 229 p->check[i] = 0 ; 230 p-> sessionId = sessionId ; 231 p->check[x] = 1 ; 232 p->count = 1 ; 233 p->cell = x ; 234 } 235 int sum = 0 ;

236 for(int i=0;i< patternNum ;i++) 237 sum +=p->check[i] ; 238 if(sum >=3) 239 p->flag = true ; 240 break ; 241 } 242 } 243 } 244 } 245 } 246 }

247 while (in2 >> IMSI >> Time) 248 { 249 string type = "" ; 250 string subType = "" ; 251 int cellId ; 252 253 in2 >> tmp ; 254 type += tmp ; 255 while (in2 >> tmp) 256 { 257 if(tmp [0] >=65 && tmp [0] <=90) 258 break ; 259 type += " " ; 260 type += tmp ;

(51)

261 } 262 subType += tmp ; 263 while (in2 >> tmp ) 264 { 265 if(tmp [0] >=48 && tmp [0] <=57) 266 break ; 267 subType += " " ; 268 subType += tmp ; 269 } 270 in2 >> cellId >> tmp >> tmp >> tmp >> tmp >> tmp; 271 hour = atoi(Time. substr (0 ,2).c_str ()) ;

272 minute = atoi(Time. substr (3 ,2).c_str ()) ; 273 second = atoi(Time. substr (6 ,2).c_str ()) ; 274 int nowTime = hour *3600+ minute *60+ second ; 275 if( subType == " Normal loc updating ")

276 {

277 if( cellId %10000 == 8440 || cellId %10000 == 9357 || cellId

%10000 == 8422 || cellId %10000 ==7175 || cellId %10000 ==3236) 278 { 279 treenode *p = Nhead ; 280 while(p != NULL) 281 { 282 if(p->IMSI > IMSI) 283 { 284 p = p->left ; 285 }

286 else if(p->IMSI < IMSI) 287 {

288 p = p->right ; 289 }

290 else 291 {

(52)

293 { 294 Timeslot tmp ; 295 tmp.begin = p->time ; 296 tmp.end = nowTime ; 297 p->T. push_back (tmp) ; 298 p->flag = true ; 299 } 300 break ; 301 } 302 } 303 } 304 } 305 }

306 while (in3 >> IMSI >> Time) 307 { 308 string type = "" ; 309 string subType = "" ; 310 int cellId ; 311 312 in3 >> tmp ; 313 type += tmp ; 314 while (in3 >> tmp) 315 { 316 if(tmp [0] >=65 && tmp [0] <=90) 317 break ; 318 type += " " ; 319 type += tmp ; 320 } 321 subType += tmp ; 322 while (in3 >> tmp ) 323 { 324 if(tmp [0] >=48 && tmp [0] <=57) 325 break ; 326 subType += " " ;

(53)

327 subType += tmp ; 328 }

329 in3 >> cellId >> tmp >> tmp >> tmp >> tmp >> tmp; 330 hour = atoi(Time. substr (0 ,2).c_str ()) ;

331 minute = atoi(Time. substr (3 ,2).c_str ()) ; 332 second = atoi(Time. substr (6 ,2).c_str ()) ; 333 int nowTime = hour *3600+ minute *60+ second ; 334 if( subType == " Normal loc updating ")

335 {

336 if( cellId %10000 == 3578 || cellId %10000 == 3642) 337 { 338 treenode *p = Shead ; 339 while(p != NULL) 340 { 341 if(p->IMSI > IMSI) 342 { 343 p = p->left ; 344 }

345 else if(p->IMSI < IMSI) 346 {

347 p = p->right ; 348 }

349 else 350 {

351 if(p->time < nowTime && nowTime -p->time <=3600)

352 { 353 Timeslot tmp ; 354 tmp.begin = p->time ; 355 tmp.end = nowTime ; 356 p->T. push_back (tmp) ; 357 p->flag = true ; 358 } 359 break ; 360 }

(54)

361 } 362 } 363 } 364 } 365 in1. close () ; 366 in2. close () ; 367 in3. close () ; 的

1 while (in >> IMSI >> Time) 2 { 3 string type = "" ; 4 string subType = "" ; 5 int RNC ; 6 int cellId ; 7 int sessionId ; 8 int finalId ; 9 10 in >> tmp ; 11 type += tmp ; 12 while (in >> tmp) 13 { 14 if(tmp [0] >=65 && tmp [0] <=90) 15 break ; 16 type += " " ; 17 type += tmp ; 18 } 19 subType += tmp ; 20 while (in >> tmp ) 21 { 22 if(tmp [0] >=48 && tmp [0] <=57) 23 break ; 24 subType += " " ; 25 subType += tmp ;

(55)

26 }

27 RNC = atoi(tmp.c_str ()) ;

28 in >> cellId >> sessionId >> finalId >> tmp >> tmp >> tmp; 29 hour = atoi(Time. substr (0 ,2).c_str ()) ;

30 minute = atoi(Time. substr (3 ,2).c_str ()) ; 31 second = atoi(Time. substr (6 ,2).c_str ()) ; 32 int nowTime = hour *3600+ minute *60+ second ; 33

34 Phone *p; 35

36 if(type == "Voice call ") 37 {

38 if(head == NULL) 39 {

40 if( finalId == 0) 41 {

42 head = new Phone( nowTime ) ; 43 head ->IMSI = IMSI ;

44 head -> cellId = cellId ;

45 head -> sessionId = sessionId ; 46 head ->call = true ;

47 head ->next = NULL ;

48 dir= search (IMSI , nowTime ) ;

49 if(( cellId == targetCell1 || cellId == targetCell2 ))

50 {

51 head ->time = nowTime ;

52 head -> deltaT = nowTime / deltaT ; 53 head ->check = 0 ;

54 N[ nowTime / deltaT ]++ ; 55 head -> through = true ; 56 head -> filter = false ;

57 }

58 } 59 }

(56)

60 else 61 {

62 bool f = false ;// flag of IMSI exists 63 p = head ;

64 while (1==1) // check IMSI exists or not 65 { 66 if(p->IMSI == IMSI) 67 { 68 f = true ; 69 if( finalId == 1) 70 { 71 if(p->call == true) 72 { 73 p->call = false ;

75 if(( cellId == targetCell1 || cellId ==

targetCell2 ) && p-> cellId == cellId )

76 { 77 p-> through = true ; 78 p-> filter = false ; 79 if(p->check == 1)// handover in 80 { 81 if(nowTime -p->time <=

limiteErlang && nowTime -p-> time >= 10)

82 {

83 if(dir == 1)

84 {

85 Nrecord [p->time/ deltaT ].

handoverIn ++ ;

86 AddErlang (p,nowTime ,1) ;

87 }

88 else if(dir == 2)

(57)

90 Srecord [p->time/ deltaT ]. handoverIn ++ ; 91 AddErlang (p,nowTime ,2) ; 92 } 93 N[p->time/ deltaT ]++ ; 94 } 95 } 96 else if(p->check == 0) 97 { 98 if(nowTime -p->time <=

limiteErlang && nowTime -p-> time >= 10) 99 { 100 if(dir == 1) 101 { 102 AddErlang (p,nowTime ,1) ; 103 } 104 else if(dir == 2) 105 { 106 AddErlang (p,nowTime ,2) ; 107 } 108 } 109 } 110 }

111 else if(( cellId != targetCell1 && cellId

!= targetCell2 ) && (p->check == 2 || p ->check == 3))// session end ouside

112 {

113 p-> through = true ; 114 p-> filter = false ;

115 if(p->check == 2 || ( p->check == 3

&& p->outtime -p->time >=10))

116 {

(58)

118 {

119 Nrecord [p-> outtime / deltaT ].

handoverOut ++ ;

120 AddErlang (p,p->outtime ,1) ;

121 }

123 {

124 Srecord [p-> outtime / deltaT ].

handoverOut ++ ; 125 AddErlang (p,p->outtime ,2) ; 126 } 127 } 128 } 129 } 130 p-> sessionId = 0 ; 131 p->check = -1 ; 132 p-> cellId = cellId ; 133 p-> filter = false ; 134 p-> through = false ; 135 p-> startime = 0 ; 136 } 137 else 138 {

139 if(p->call == false) // new call

140 {

141 p->call = true ;

142 p-> sessionId = sessionId ; 143 p-> startime = nowTime ; 144 dir= search (IMSI , nowTime ) ;

targetCell2 ))

146 {

147 p->time = nowTime ;

(59)

149 p->check = 0 ; 150 N[p->time/ deltaT ]++ ; 151 p-> through = true ; 152 p-> filter = false ; 153 } 154 } 155 else // handover 156 {

targetCell2 ) && (p-> cellId != targetCell1 && p-> cellId != targetCell2 ))// handoverIn

159 {

160 if(p->check == 2 && nowTime -p->

outtime < HItime )

161 {

162 p->check = 1 ;

163 }

164 else if(p->check == 3 && nowTime -p->

outtime < HItime )

165 {

166 p->check = 0 ;

167 }

168 else if((p->check == 2 || p->check ==

3) && nowTime -p->outtime >= HItime )

169 {

170 if(p->check == 2 || ( p->check ==

3 && p->outtime -p->time >=10) )

171 {

172 if(dir == 1)

(60)

174 AddErlang (p,p->outtime ,1)

;

175 Nrecord [p-> outtime / deltaT

]. handoverOut ++ ; 176 } 177 else if(dir == 2) 178 { 179 AddErlang (p,p->outtime ,2) ;

180 Srecord [p-> outtime / deltaT

]. handoverOut ++ ;

181 }

182 }

184 p-> deltaT = nowTime / deltaT ; 185 p-> outtime = 0 ; 186 p->check = 1 ; 187 p-> sessionId = sessionId ; 188 } 189 else 190 { 191 p->time = nowTime ;

192 p-> deltaT = nowTime / deltaT ; 193 p->check = 1 ;

194 }

195 }

196 else if((p-> cellId == targetCell1 || p->

cellId == targetCell2 ) && ( cellId != targetCell1 && cellId != targetCell2 )) // handoverOut

197 {

198 if( cellId %10000 != p-> cellId %10000) //

leave this BS set

(61)

200 if(p->check == 0)

201 {

202 if(nowTime -p->time <=

limiteErlang && nowTime -p ->time >= 10) 203 { 204 if(dir == 1) 205 { 206 AddErlang (p,nowTime ,1) ; 207 Nrecord [ nowTime / deltaT ]. handoverOut ++ ; 208 } 209 else if(dir == 2) 210 { 211 AddErlang (p,nowTime ,2) ; 212 Srecord [ nowTime / deltaT ]. handoverOut ++ ; 213 } 214 } 215 p->check = -1 ; 216 p->time = 0 ; 217 p-> deltaT = 0 ; 218 p-> sessionId = sessionId ; 219 }

220 else if(p>check == 1 && nowTime

-p->time >= HItime )

221 {

222 if(nowTime -p->time <=

limiteErlang )

(62)

224 p-> through = true ; 225 p-> filter = false ; 226 if(dir == 1) 227 { 228 AddErlang (p,nowTime ,1) ; 229 Nrecord [p->time/ deltaT ]. handoverIn ++ ; 230 Nrecord [ nowTime / deltaT ]. handoverOut ++ ; 231 } 232 else if(dir == 2) 233 { 234 AddErlang (p,nowTime ,2) ; 235 Srecord [p->time/ deltaT ]. handoverIn ++ ; 236 Srecord [ nowTime / deltaT ]. handoverOut ++ ; 237 } 238 N[p->time/ deltaT ]++ ; 239 } 240 p->check = -1 ; 241 p->time = 0 ; 242 p-> deltaT = 0 ; 243 p-> sessionId = sessionId ; 244 } 245 }

246 else if(nowTime -p->time >= HOtime )

(63)

248 if(p->check == 0) 249 { 250 if(nowTime -p->time <= limiteErlang ) 251 { 252 if(dir == 1) 253 { 254 AddErlang (p,nowTime ,1) ; 255 Nrecord [ nowTime / deltaT ]. handoverOut ++ ; 256 } 257 else if(dir == 2) 258 { 259 AddErlang (p,nowTime ,2) ; 260 Srecord [ nowTime / deltaT ]. handoverOut ++ ; 261 } 262 } 263 p->check = -1 ; 264 p->time = 0 ; 265 p-> deltaT = 0 ; 266 p-> sessionId = sessionId ; 267 }

268 else if(p>check == 1 && nowTime

-p->time >= HItime ) 269 { 270 if(nowTime -p->time <= limiteErlang ) 271 { 272 p-> through = true ;

(64)

273 p-> filter = false ; 274 if(dir == 1) 275 { 276 AddErlang (p,nowTime ,1) ; 277 Nrecord [p->time/ deltaT ]. handoverIn ++ ; 278 Nrecord [ nowTime / deltaT ]. handoverOut ++ ; 279 } 280 else if(dir == 2) 281 { 282 AddErlang (p,nowTime ,2) ; 283 Srecord [p->time/ deltaT ]. handoverIn ++ ; 284 Srecord [ nowTime / deltaT ]. handoverOut ++ ; 285 } 286 N[p->time/ deltaT ]++ ; 287 } 288 p->check = -1 ; 289 p->time = 0 ; 290 p-> deltaT = 0 ; 291 p-> sessionId = sessionId ; 292 } 293 }

time < HItime )

(65)

296 p->check = -1 ; 297 p->time = 0 ; 298 p-> outtime = 0 ; 299 p-> deltaT = 0 ;

300 }

time >= HItime )

302 {

303 if(dir ==1)

304 {

305 Nrecord [p->time/ deltaT ].

handoverIn ++ ;

306 }

308 {

309 Srecord [p->time/ deltaT ].

handoverIn ++ ; 310 } 311 N[p->time/ deltaT ]++ ; 312 p-> through = true ; 313 p-> filter = false ; 314 p->check = 2 ; 315 p-> outtime = nowTime ; 316 } 317 else if(p->check == 0) 318 { 319 p->check = 3 ; 320 p-> outtime = nowTime ; 321 } 322 }

323 else if((p->check == 2 || p->check == 3))

324 {

325 if(p->check == 2 || ( p->check == 3

(66)

326 {

327 if(dir == 1)

328 {

329 AddErlang (p,p->outtime ,1) ; 330 Nrecord [p-> outtime / deltaT ].

handoverOut ++ ;

331 }

333 {

334 AddErlang (p,p->outtime ,2) ; 335 Srecord [p-> outtime / deltaT ].

handoverOut ++ ; 336 } 337 } 338 p->check = -1 ; 339 p->time = 0 ; 340 p-> deltaT = 0 ; 341 p-> sessionId = sessionId ; 342 } 343 } 344 p-> cellId = cellId ; 345 p-> sessionId = sessionId ; 346 } 347 break ; 348 } 349 if(p->next == NULL) 350 break ; 351 p = p->next ; 352 } 353 if(f == false) 354 { 355 if( finalId == 0) 356 {

(67)

358 p = p->next ; 359 p->IMSI = IMSI ; 360 p-> cellId = cellId ; 361 p-> sessionId = sessionId ; 362 p->call = true ; 363 p->next = NULL ;

targetCell2 ))

366 {

368 p-> deltaT = nowTime / deltaT ; 369 p->check = 0 ; 370 N[p->time/ deltaT ]++ ; 371 p-> through = true ; 372 p-> filter = false ; 373 } 374 } 375 } 376 } 377 } 378 }

基於行動電信網路的車速量測之研究

國

立

交

通

大

學

網路工程研究所

碩

士

論

文

基於行動電信網路的車速量測之研究

Deriving the Vehicle Speeds from Mobile

Telecommunications Network

研 究 生：張裕隆

指導教授：林一平 教授

基 於 行 動 電 信 網 路 的 車 速 量 測 之 研 究

Deriving the Vehicle Speeds from Mobile Telecommunication Network

研 究 生：張裕隆 Student：Yu-Long Chang

指導教授：林一平 Advisor：Yi-Bing Lin

國 立 交 通 大 學

網 路 工 程 研 究 所

碩 士 論 文

A Thesis

Submitted to Institute of Network Engineering

College of Computer Science

National Chiao Tung University

in partial Fulfillment of the Requirements

for the Degree of

Master

in

Computer Science

June 2013

Hsinchu, Taiwan, Republic of China

基於行動電信網路的車速量測之研究

網路

研究

Deriving the Vehicle Speeds from Mobile

Telecommunications Network

Student: Yu-Long Chang

Advisor: Dr. Yi-Bing Lin

Institute of Network Engineering

National Chiao Tung University

ABSTRACT

誌

謝

Cell i-1

Cell i-2

Cell i

Cell i+1

Cell i+2

Road

Internet

Mobile Core Network

Network

Probe

ITS Server

b

c

d

f

g

h

a

MSC

e

Lin-Chang-Huangfu (LCH)

的

(Bias)

time

車速

之

g

R

oa

d 1

Cell

i

Cell

研究生：張裕隆

指導教授：林一平教授

基於行動電信網路的車速量測之研究

研究生：張裕隆 Student：Yu-Long Chang

國立交通大學

網路工程研究所

碩士論文

_n