Professional Documents
Culture Documents
INTERNET
One Gateway
y
1 2 x n-1 n
1
H ( x) [( x 1) ( x 2) ... 1 0 1 2 ... (n x)]
n
Average number of hops
1 n 2 1
2
n 1
H ( x) x
n 2 4
To minimize H(x), we have
n 1 n 1
x orx
2 2
One Gateway
1 2 3 4 5 6 7 8 9
n 1
n9 x 5
2
1 2 3 4 5 6 7 8
n 1
n 8 x 4
2
Multiple
p Gateways
y
How to select m gateways from n APs?
n
x1
x2
xm
1 n
n1 n2 nm
1 ni 1
2
ni2 1
H i ( xi ) xi
ni 2 4
A
Average number
b off hhops off each
h group.
Multiple Gateways
n
Let C be the number of APs in a group
m
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
5 1
x1
2
Multiple Gateways
n
Let C be the number of APs in a group
m
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
4
x1
2
Multiple Gateways
n
Let C be the number of APs in a group
m
Case3: C is not an integer but exists k(1km) such that C=(n-k)/m is an odd integer :
C 1
(i 1)C , if 1 i m k 1
xi 2
xm k 1 [i (m k 1)](C 1) , if m k 2 i m
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
C C C+1 C+1
Multiple Gateways
n
Let C be the number of APs in a group
m
Case3: C is not an integer but exists k(1km) such that C=(n-k)/m is an even integer :
C
(i 1)C , if 1 i m k
xi 2
xm k [i (m k )](C 1) , if m k 1 i m
C C C C+1 C+1
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
C C C+1 C+1
Energy Balance
GatewayAP
Energy Balance
Pt transmitted power
Pr received power
Gt antenna gains for the transmitter
Pt Gt Gr
2
Pr Gr antenna gains for the receiver
(4 ) 2 Ld
L( L 1) the system loss factor
the wavelength
g
the path loss exponent
Energy Balance
Given n APs and the gateway has been known,
how to determine the distance of neighboring APs for achieving energy balance?
Pt Gt Gr 2 (4 ) 2 L
Pr P P d Td
(4 ) 2 Ld Gt Gr 2
t r
1 2 x-1 x n-x n-1 n
GatewayAP
y AP !
One Gateway
x 1 n x
P id i jd n j T
i 1 j 1
One Gateway
x 1 n x
i n j
id jd
1
x 1 n x n 1
id i jd n j
i 1 j 1
n 1 i 1 j 1
1
x 1 n x n 1
P (n 1) id i jd n j T
i 1 j 1
x 1 n x
Since d d
i 1
i
j 1
j L
i 1 i i x ( n i )
x 1 1 n 1 1
1
L , for i 1
1
d i i d1 , for i [ 2, x 1]
1
( n i ) d1 , for i [ x, n 1]
GatewayAP
y AP !
Multiple Gateways
1 x1
xi
xm n
S1 t1 Si ti Sm tm
P {[d1 2d 2 ... ( x1 1)d x1 1 ] [d t1 1 2d t1 2 ... (t1 x1 )d x1 ]}T
{[d t1 1 2d t1 2 ... ( x2 1)d x2 1 ] [d t2 1 2d t2 2 ... (t 2 x2 )d x2 ]}T
...
{[d tm1 1 2d tm1 2 ... ( xm 1)d xm 1 ] [d n1 2d n 2 ... (n xm )d xm ]}T
m xk 1 t k xk
P id tk 1 i jd tk j T
k 1 i 1 j 1
Multiple Gateways
1
xk 1 tk xk
m n 1
P (n 1) id tk 1 i jd tk j T
k 1 i 1
j 1
m xk 1 t k xk
Since d t k 1 i d t k j L
k 1 i 1 j 1
k 1 i 1 i j xk (t k j )
m x k 1 1 t k xk 1
1
L , for i t k 1
1
d i i d1 , for i [t k 2, xk 1 1]
1
(t k 1 i ) d1 , for i [ xk 1 , t k 1 1]
2-D Vehicular Networks
INTERNET
One Gateway
g ( x, y ) denotes the location of the ggatewayy
( x1 , y1 ) ( xi , yi ) denotes the location of AP i
r ( n ) denotes the common transmission range for all n APs
( x x1 ) 2 ( y y1 ) 2
( x x3 ) 2 ( y y3 ) 2
To minimize H(g), we have
n
Minimize f ( x, y ) ( x xi ) 2 ( y yi ) 2
( x x2 ) 2 ( y y 2 ) 2
( x3 , y3 ) i 1
( x2 , y 2 )
One Gateway
1. Initialization:
1 n
( x1 , y1 ) x xi
n i 1
1 n Not the ideal location
y i 1 yi
n
2. Iteration: calculate Scos and Ssin, where
n
( xi x)
Ideal location Scos
i 1 ( x xi ) 2 ( y yi ) 2
n
( yi y )
Initialization location Ssin
i 1 ( x xi ) 2 ( y yi ) 2
( x, y ) 3. IF |Scos|Threshold and |Ssin|Threshold
4. Go to 9.
5. Else
6. x=x+step*Scos and y=y+step*Ssin
7. Go to 2.
( x2 , y 2 ) ( x3 , y 3 ) 8. END IF
9. Finish
One Gateways
Multiple Gateways
INTERNET
Discussion-Power
P Pt H (G )
Pt is the transmission power, which is the same for all APs,
G is the gateways in the network
H(G) is the average number of hops from an AP to a gateway
H (G ) n C
denotes the AP generates traffics.
C denotes the capacity of the network.
H(G) is the average number of hops from an AP to a gateway.
n is the total number of APs in the network.
1 2 3 4 5
1 2 3 4 5
H(G) = 2 nH(G) = 10
= 6 Mbps
RSU
RSU
Introduction The Architecture of Vehicle (Roadside service)
Performance
f Metrics
i (Tradeoff )
Service Ratio ()
Ratio of the number of requests served before the service
deadline to the total number of arriving requests.
D t Q
Data Quality
lit ()
Percentage of fresh data access
System Model
Each request is characterized by
v-id : the Vehicle-ID
d-id : the Data-ID
op : upload / download
deadline : Time Constraint
FDF
SDF
Scheduling Scheme I
D
D*S
S Scheduling
D*S Scheduling
D:D D-List
Li t (Deadline-list)
(D dli li t)
S : S-List (DataSize-list)
D-List S-List
Current Clock = 100
(Deadline) (Data Size)
a 101 a 50 e
b 102 b 100 c
c 105 c 150 b
Sort by Sort by
d 106 Deadline d Data Size 200 f
e 110 e 350 a
f 115 f 500 d
g 116 g 1000 g
Requests in RSU
Scheduling Scheme I
D
D*S
S Scheduling
Basic Idea
Given two requests with the same deadline, the
one asking for a small size data should be served
first.
Given two requests asking for the data items with
same size, the one with an earlier deadline should
be served first.
Assign each arrival request a service value based
on its deadline and data size,
size called DS_value
DS value as
its service priority weight
DS value = (Deadline CurrentClock) * DataSize
DS_value
Scheduling Scheme I
D
D*S
S Scheduling Implementation
Information in RSU
D-List S-List
(Deadline) (Data Size)
a 101 50 e
b 102 100 c
c 105 150 b
d 106 200 f
e 110 350 a
f 115 500 d
g 116 1000 g
Current Clock = 100
Scheduling Scheme I
D
D*S
S Scheduling Implementation
MinDS Step 1
= (Deadline - CurrentClock) * DataSize
Current Clock = 100
= (101-100) * 350 = 350
DS_Value
DS Value 350
D-List S-List
D-List entry request b
(Deadline) (Data Size)
request b DS_Value
a 101 50 e DataSize MinS
c 105 150 b
MinS = 175
d 106 200 f
e 110 350 a
f 115 500 d
g 116 1000 g
Scheduling Scheme I
D
D*S
S Scheduling Implementation
MinDS Step 1
= (Deadline - CurrentClock) * DataSize
= (101-100) * 350 = 350
D-List S-List
(Deadline) (Data Size) Current Clock = 100
checked a 101 50 e
c 105 150 b
MinS = 175
d 106 200 f
e 110 350 a
f 115 500 d
g 116 1000 g
Scheduling Scheme I
D*S Scheduling Implementation
Step 2
Current Clock = 100
DS_Value (request e)
DS_Value 350 = (Deadline - CurrentClock) * DataSize
SS-List
List entry request cc = (110-100)
(110 100) * 50 = 500 > MinDS = 350
request c DS_Value
Deadline MinD D-List S-List
350 > ( MinD
in 100
00 ) * 100
00 (Deadline) (Data Size)
b 102 100 c
Mi D = 103.5
MinD 103 5
c 105 150 b
d 106 200 f
e 110 350 a
f 115 500 d
g 116 1000 g
Scheduling Scheme I
D
D*S
S Scheduling Implementation
Step 2
DS_Value (request e)
= (Deadline - CurrentClock) * DataSize
= (110-100)
(110 100) * 50 = 500 > MinDS = 350
D-List S-List
Current Clock = 100 (Deadline) (Data Size)
a 101 50 e
b 102 100 c
Mi D = 103.5
MinD 103 5
c 105 150 b
d 106 200 f
e 110 350 a
f 115 500 d
g 116 1000 g
Scheduling Scheme II
Download Optimization: Broadcasting
Observation
several requests may ask for downloading the same
data item
wireless communication is broadcast in nature
Basic Idea
delay some requested data and broadcast it before
the deadlines,, then several requests
q mayy be served
via a single broadcast
the data with more pending requests should be
servedd first
fi t
DSN_value = ( Deadline CurrentClock ) * DataSize/Number
Basic Idea
two priority queues: one for the update requests and
the other for the download requests.
the data server provides two queues with different
bandwidth (i.e., service probability)
Simulation Setup
NS-2
NS 2
400m*400m square street scenario
One RSU server is located at the center of two 2-way y roads
40 vehicles randomly deployed on each lane
Each vehicle issues request with a probability p
Access pattern of each data item follows Zipf distribution
Effect of Workload
As workload
A kl d iincreases, D*S/N can achieve
hi the
th hi
highest
h t
service ratio while its data quality degrades dramatically
Effect of Workload
As workload
A kl d iincreases, D*S/N can achieve
hi the
th hi
highest
h t
service ratio while its data quality degrades dramatically
Effect of Workload
As workload
A kl d iincreases, D*S/N can achieve
hi the
th hi
highest
h t
service ratio while its data quality degrades dramatically
C l i
Conclusion
This paper proposed a basic scheduling scheme called D*S to
consider both service deadline and data size when making
scheduling
h d li decisions.
d ii
To make use of the wireless broadcasting, this paper proposed
a new scheduling
h d li scheme
h called
ll d D*S/N tot serve multiple
lti l
requests with a single broadcast.
This paper also identified the effects of upload requests on
data quality, and proposed a Two-Step scheduling scheme to
pprovide a balance between servingg download and update
p
requests.