Application scenario WMNs offer a promising networking architecture to provide multimedia services...

18
Application scenario WMNs offer a promising networking architecture to provide multimedia services to mobile users WMNs represent an attractive solution to extend the Internet access over local areas and metropolitan areas PROBLEM The spectrum resource available POSSIBLE SOLUTION: Use the Cognitive Radio paradigm FRAMEWORK: Consider Active Mesh Networks (Content- aware Cognitive Wireless Mesh Network s)
  • date post

    19-Dec-2015
  • Category

    Documents

  • view

    213
  • download

    0

Transcript of Application scenario WMNs offer a promising networking architecture to provide multimedia services...

Application scenario• WMNs offer a promising

networking architecture to provide multimedia services to mobile users

• WMNs represent an attractive solution to extend the Internet access over local areas and metropolitan areas

• PROBLEM▫ The spectrum resource

available• POSSIBLE SOLUTION:

▫ Use the Cognitive Radio paradigm

• FRAMEWORK:▫ Consider Active Mesh

Networks (Content-aware Cognitive Wireless Mesh Network s)

The envisioned Active Mesh Net architecture

Internet

Gateway

MR33

C3.3

C3.2

C3.1

f3

f3

f3

f3

MR23

C2.2

C2.1

f2

f2

f2

MR13

C1.1

C1.2C1.3

C1.4

f1

f1

f1

f1

f1

• Formed by interconnecting several cluster of mobile Mesh Clients (MCs) via a wireless backbone composed by static Mesh Router (MRs)

• Dowlink and uplink traffic

• Each MR acts as access point

• Frequencies {fi ,i=1,2,3} are

used both to receive data from the MC(j) by MR(i+1)

Cluster 3

Cluster 1

Cluster 2

Cognitive functionality

• MCs are battery-powered• Fading affecting the wireless link, between MCs and MR, is

assumed constant over each slot (block fading)• MC carry out Channel Detection and Channel Estimation• MR carry out Belief propagation and Soft Data Fusion

MC i,1MC i,2

MC i,3

MR(i) (access point)

Cluster i-th

MR(i+1) (access point)

• MC problem:▫ Optimal access rate

and flow-control• MR problem:

▫ Optimal set of the access times

fifi

fi

fi

Clients’Payloa

dACKvv

Belief Propagatio

n

Soft Data

Fusion

ChannelEstinatio

n

Resource Allocation

andClient’s

scheduling

ChannelDetection

Intra-cluster slot structure

• Slot-duration of TS (sec.)• It is split into Lt minislot• Each MC(j) uses:

▫ LD minislot for Channel Detection phase▫ LE minislot for Channel Estimation phase▫ LP minislot to transmit data to MR(i)▫ LA minislot to receive Ack message

• Each MR(i) uses:▫ LB minislot for Belief Propagation phase▫ LFminislot for Soft Data Fusion phase▫ LA minislot to sent Ack message

• MR(i) and MC(j) use:▫ LS minislot for Resource Allocation and Clients’

Scheduling

LD LB LF LE LS LP LA

Channel Learning

• MCs are listening to the channel

Channel Detection

Clients’Payload

vvChannelEstimatio

n

ChannelDetection

MC functionalities

1

( ; ) ( ; ) if ( ) 0

( ; ) ( , ) ( ; ) ( ; ) if ( ) 1

j j i

j ji i j i

k t n k t a t

k t g k t s k t n k t a t

Sample (deterministic or aleatory) generated by MR(i+1) in the minislot k-th

Channel Coefficient MR(i+1)-MC(j) in the minislot k-th

State of primary user’s activity

LD LE LP

Channel Estimation

• MR(i) transmits a known pilot’s sequence• MC(j) known this sequence• MC(j) calculates the channel estimation based on:

( ) ( ) ( , )ji ji jt h t p v k t

Pilot sequence a priori know

Noise sequence

Clients’Payload

vvChannel

EstimationChannel

DetectionMC functionalities

LD LE LP

Channel Estimation

• MR(i) transmits a known pilot’s sequence• MC(j) known this sequence• MC(j) calculates the channel estimation based on:

• Each MC use these minislots to transmit data to MR

( ) ( ) ( , )ji ji jt h t p v k t

Pilot sequence a priori know

Noise sequence

Clients’Payload

vvChannel

EstimationChannel

DetectionMC functionalities

LD LE LP

Belief Propagation

• Definition:▫ At the beginning of each slot, each access point MR(i)

estimates and/or updating the following conditional probability

ACKBelief

Propagation

Soft Data

FusionMR functionalities

LB LF LA

Belief Propagation

• Definition:▫ At the beginning of each slot, each access point MR(i)

estimates and/or updating the following conditional probability

ACKBelief

Propagation

Soft Data

Fusion

( ( ) 1| ( ))i iP a t t Set of the informations

about the MR(i+1) activity in the previous slot (t-1)

MR functionalities

LB LF LA

Belief Propagation

• Definition:▫ At the beginning of each slot, each access point MR(i)

estimates and/or updating the following conditional probability

• Noncooperative: when is empty set or contains informations about only the MR(i+1) of the cluster i-th

• Cooperative: when is nonempty and it contains informations about the previous activities all primary users

ACKBelief

Propagation

Soft Data

Fusion

( ( ) 1| ( ))i iP a t t Set of the informations

about the MR(i+1) activity in the previous slot (t-1)

( )i t

( )i t

MR functionalities

LB LF LA

Data Fusion (1/3)

• Each MR(i) knows the primary’s activity only at the end of the slot t-th but MR(i) must know the state of MR(i+1) at the beginning of the phase Resource Allocation

1. MR(i) merges (Data Fusion) decisions already calculated by MC(j) in the first part of Channel Detection

2. MR(i) calculates a posteriori probabilities that the i-th channel is transmission free

ACKBelief

Propagation

Soft Data

FusionMR functionalities

LB LF LA

Data Fusion (2/3)

• Definition:▫ Algorithm that computes the conditional

probability. This last is computed by each MR(i) as in

MR functionalitiesACKBelief

Propagation

Soft Data

Fusion

LB LF LA

Data Fusion (2/3)

• Definition:▫ Algorithm that computes the conditional

probability. This last is computed by each MR(i) as in

( ) ( ( ) 0 | ( ))iL i iP t P a t V t

Set of the informations about the MR(i+1) activity. This informations are available at the end of the Channel Detection

phase

MR functionalitiesACKBelief

Propagation

Soft Data

Fusion

LB LF LA

Data Fusion(3/3)

( )

( ) ( )

( ( ) 0 | ( ))[ ( ( ) | ( ) 0)]

( )( ( ) 0 | ( ))[ ( ( ) | ( ) 0)] ( ( ) 1| ( ))[ ( ( ) | ( ) 1)

i

i i

i i ji ij C ti

Li i ji i i i ji i

j C t j C t

P a t t P a t a t

P tP a t t P a t a t P a t t P a t a t

( ) {{ ( ), ( )}, ( )}, i , 1i ji i iV t a t j C t t I t

Set of the MCs belonging to i-th cluster

Number of clusters

Optimal Soft Data Fusion

Data Fusion(3/3)

( )

( ) ( )

( ( ) 0 | ( ))[ ( ( ) | ( ) 0)]

( )( ( ) 0 | ( ))[ ( ( ) | ( ) 0)] ( ( ) 1| ( ))[ ( ( ) | ( ) 1)

i

i i

i i ji ij C ti

Li i ji i i i ji i

j C t j C t

P a t t P a t a t

P tP a t t P a t a t P a t t P a t a t

( ) {{ ( ), ( )}, ( )}, i , 1i ji i iV t a t j C t t I t

Set of the MCs belonging to i-th cluster

Number of clusters

Optimal Soft Data Fusion

( )iLP t• represents the conditional probability that the i-th

channel is available

( ( ) | ( ))i iP a t t• MR(i) knows probability from the Belief Propagation phase

Hard or Soft Data Fusion?

• Hard Data Fusion▫ MCs provide hard informations (i.e., binary decisions)

to the corresponding MR▫ MR provides hard informations

• Soft Data Fusion▫ MCs provide the observations directly to the MR ▫ MR processes the set of the observations▫ MR provides hard decisions

• My Data Fusion? Hard or Soft?• Neither hard nor soft

▫ MCs provide the soft informations (in form of Probability) to the MR

▫ MR processes the soft informations▫ MR provides a soft information (in form of Probability)

P.K.Varshney, ‘Distributed Detection and Data Fusion’, Springer, 1997

ACK

• MR(i) sent an Ack message defined in the following as:

• MC(j) receive ‘zero’, in that case:▫ MR(i+1) was not active in that slot▫ MC(j) removed from the queue the IUs that it has

transmitted in the slot t-th

• MC(j) receive ‘one’ , in that case:▫ M(i+1) was active in that slot▫ MC(j) not remove the IUs

( ) 1 ( )iA iZ t a t

MR functionalitiesACKBelief

Propagation

Soft Data Fusion

LB LALF

Binary variablethat defines Ack

message

Work in Progress

•Develop in closed-form expressions for the optimal access rate and the optimal access time

•Unconditional optimization problem

•Performance evaluation of the overall Active Mesh architecture