Minimizing interference for the highway model in Wireless Ad-hoc and Sensor Networks Haisheng Tan, Tiancheng, Lou, Francis C.M. Lau, YuexuanWang, Shiteng Chen CS, The University of Hong Kong, Hong Kong, China ITCS, Tsinghua University, Beijing, China Jan. 25 th , SOFSEM, 2011
54
Embed
Minimizing interference for the highway model in Wireless Ad-hoc and Sensor Networks Haisheng Tan, Tiancheng, Lou, Francis C.M. Lau, YuexuanWang, Shiteng.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Minimizing interference for the highway model in Wireless Ad-hoc and Sensor Networks
Haisheng Tan, Tiancheng, Lou, Francis C.M. Lau, YuexuanWang, Shiteng Chen
CS, The University of Hong Kong, Hong Kong, ChinaITCS, Tsinghua University, Beijing, China
Jan. 25th, SOFSEM, 2011
Outline
IntroductionProblem DefinitionsMinimizing the Average InterferenceMinimizing the Maximum InterferenceDiscussions and Future workQ & A
Introduction
Wireless Ad hoc and Sensor Networks
Introduction
Wireless Ad hoc and Sensor Networks
Environmental monitoring, intrusion detection, health care, etc.
Smart Earth (IBM), Sense China …
Introduction
Energy !
Introduction
Energy !Interference
Introduction
Energy !Interference
Receiver-centric interference transmission radius of u
Problem Definitions
the average interference of a graph G
the maximum interference of a graph G
Problem Definitions
the average interference of a graph G
the maximum interference of a graph G
Problems: Given nodes arbitrarily deployed along a 1D line (the highway
model) Connected Min-Avg or Min-max interference The optimal solution is actually a spanning tree.
Observations
Observations
small node degrees
Observations
small node degreessparse topology
Observations
small node degreessparse topologyNearest Neighbor Forest (each node is connected
to its nearest neighbor)
Observations
small node degrees sparse topology Nearest Neighbor Forest (each node is connected
to its nearest neighbor)
a)
b)
c)
Minimizing the Average Interference
In 2D networks: an asymptotically optimal algorithm with an approximation ratio of O(log n) (Moscibroda et al. 2005)
Minimizing the Average Interference
In 2D networks: an asymptotically optimal algorithm with an approximation ratio of O(log n) (Moscibroda et al. 2005)
In the highway model (Our work):
a polynomial-time exact algorithm
Minimizing the Average Interference
In 2D networks: an asymptotically optimal algorithm with an approximation ratio of O(log n) (Moscibroda et al. 2005)
In the highway model (Our work):
1. No-cross property
Minimizing the Average Interference
In 2D networks: an asymptotically optimal algorithm with an approximation ratio of O(log n) (Moscibroda et al. 2005)
In the highway model (Our work):
1. No-cross property when |ac| <=|
bc|+|cd|
Minimizing the Average Interference
In the highway model: 2. Calculate the total interference via the interference created
by each node
Minimizing the Average Interference
In the highway model: 2. Calculate the total interference via the interference created
by each node
Minimizing the Average Interference
In the highway model: 2. Calculate the total interference via the interference created
by each node
Independent sub-problems
Minimizing the Average Interference
Two questions: How to divide the whole line into sub-segments How to connect the nodes inside each segment
Minimizing the Average Interference
Two questions: How to divide the whole line into sub-segments How to connect the nodes inside each segment
Functions for DP F(s,t), s<t, which is short for Compute the minimum total interference created by the
nodes from s+1 to t-1 , such that
Minimizing the Average Interference
Two questions: How to divide the whole line into sub-segments How to connect the nodes inside each segment
Functions for DP F(s,t), s<t, which is short for
OR
Minimizing the Average Interference
Two questions: How to divide the whole line into sub-segments How to connect the nodes inside each segment
Functions for DP F(s,t), s<t, which is short for
OR
Minimizing the Average Interference
Functions for DP G(s,t), s<t Compute the minimum total interference created by nodes
from s +1 to t-1, such that
Minimizing the Average Interference
Functions for DP G(s,t), s<t
Minimizing the Average Interference
Functions for DP G(s,t), s<t
Minimizing the Average Interference
Functions for DP G(s,t), s<t
The minimum average interference
Minimizing the Average Interference
Correctness Verified by the brute-force search running in time
the maximum node degree
Minimizing the Average Interference
Correctness Verified by the brute-force search running in time
Time complexity:
the maximum node degree
Minimizing the Average Interference
Correctness Verified by the brute-force search running in time
Time complexity:
(the numbers are the interference created by the nodes)
the maximum node degree
Minimizing the Average Interference
Correctness Verified by the brute-force search running in time
Time complexity:
(the numbers are the interference created by the nodes)
Can we do better ?? Y!
the maximum node degree
Minimizing the Maximum Interference
Harder!! No-cross property: still holds
Minimizing the Maximum Interference
Harder!! No-cross property: still holds Independent sub-segments: not found
Minimizing the Maximum Interference
Harder!! No-cross property: still holds Independent sub-segments: not found
In 2D networks: NP-hard (Buchin 2008) Bounded in
Minimizing the Maximum Interference
Harder!! No-cross property: still holds Independent sub-segments: not found
In 2D networks: NP-hard (Buchin 2008) Bounded in In 1D networks: An appr. with ratio (von Richenbach, et al. 2005) A sub-exponential-time exact algorithm (Our work)
Minimizing the Maximum Interference
Check whether the min-max can be k, where 1<k<n
Minimizing the Maximum Interference
Check whether the min-max can be k, where 1<k<n
A skeleton : Record the nodes from s to t that can interfere with nodes
outside [s,t] with their transmission radii
Minimizing the Maximum Interference
Check whether the min-max can be k, where 1<k<n
A skeleton : Record the nodes from s to t that can interfere with nodes
outside [s,t] with their transmission radii
Minimizing the Maximum Interference
Check whether the min-max can be k, where 1<k<n
A skeleton : Record the nodes from s to t that can interfere with nodes
outside [s,t] with their transmission radii
Minimizing the Maximum Interference
Functions: boolean F*(s,t), which is short for
Minimizing the Maximum Interference
Functions: boolean F*(s,t), which is short for
OR
Minimizing the Maximum Interference
Functions: boolean F*(s,t), which is short for
OR
Minimizing the Maximum Interference
Functions: boolean G*(s,t)
Minimizing the Maximum Interference
Functions: boolean G*(s,t)
Minimizing the Maximum Interference
Functions: boolean G*(s,t)
Minimizing the Maximum Interference
Functions: boolean G*(s,t)
Check the whole line
Minimizing the Maximum Interference
Time complexity # of the different valid skeletons for a segment from s to t,
where s>0 and t<n-1:
Minimizing the Maximum Interference
Time complexity # of the different valid skeletons for a segment from s to t,
where s>0 and t<n-1:
Time complexity:
Minimizing the Maximum Interference
Time complexity # of the different valid skeletons for a segment from s to t,
where s>0 and t<n-1:
Time complexity:
Can we do better? No idea yet
Discussion and Future work
PlanarityMultiple optimal spanning trees
the min-max for the 6-node exponential chain
Discussion and Future work
PlanarityMultiple optimal spanning trees
Is min-max in 1D NP-hard? How about 3D networks?How to design efficient approximations to minimize
the maximum in 2D networks?How to tackle interference minimization with other
network properties, such as small node degree and spanner?