+ All Categories
Home > Documents > Http://dblab.usc.edu Announcements Your homework is due on September 19 th. Your homework is due on...

Http://dblab.usc.edu Announcements Your homework is due on September 19 th. Your homework is due on...

Date post: 21-Dec-2015
Category:
View: 213 times
Download: 0 times
Share this document with a friend
Popular Tags:
25
http:// http:// dblab.usc.edu dblab.usc.edu Announcements Announcements Your homework is due on September Your homework is due on September 19 19 th th . . I will be away starting Sept 5 I will be away starting Sept 5 th th and and will see you on Sept 19 will see you on Sept 19 th th . . Make sure to attend lectures and listen Make sure to attend lectures and listen to video taped tutorials. They are to video taped tutorials. They are essential for your homework and essential for your homework and project. project. Professor Francois Bar is guest Professor Francois Bar is guest lecturer on September 26 lecturer on September 26 th th . . Professor Anne Friedberg is guest Professor Anne Friedberg is guest lecturer on October 10 lecturer on October 10 th th . .
Transcript

http://dblab.usc.eduhttp://dblab.usc.edu

AnnouncementsAnnouncements Your homework is due on September 19Your homework is due on September 19thth.. I will be away starting Sept 5I will be away starting Sept 5thth and will see and will see

you on Sept 19you on Sept 19thth.. Make sure to attend lectures and listen to video Make sure to attend lectures and listen to video

taped tutorials. They are essential for your taped tutorials. They are essential for your homework and project.homework and project.

Professor Francois Bar is guest lecturer on Professor Francois Bar is guest lecturer on September 26September 26thth. .

Professor Anne Friedberg is guest lecturer Professor Anne Friedberg is guest lecturer on October 10on October 10thth..

http://dblab.usc.eduhttp://dblab.usc.edu

Any Homework Questions?Any Homework Questions?

http://dblab.usc.eduhttp://dblab.usc.edu

Question from last weekQuestion from last week

CSCI 599: Replication in Ad CSCI 599: Replication in Ad Hoc Network of Wireless Hoc Network of Wireless

DevicesDevices

Shahram GhandeharizadehShahram GhandeharizadehComputer Science DepartmentComputer Science Department

University of Southern CaliforniaUniversity of Southern California

http://dblab.usc.eduhttp://dblab.usc.edu

OutlineOutline Target environment: H2O networks.Target environment: H2O networks. Definition of access cost.Definition of access cost. GreedyDual replacement technique.GreedyDual replacement technique. Conclusion and future research.Conclusion and future research.

http://dblab.usc.eduhttp://dblab.usc.edu

Target EnvironmentTarget Environment Key characteristics:Key characteristics:

Limited by the bandwidth and radio range of Limited by the bandwidth and radio range of wireless devices (less constrained by energy and wireless devices (less constrained by energy and mobility).mobility).

Management of devices is decentralized.Management of devices is decentralized. Each device is autonomous.Each device is autonomous. Devices cooperate when in radio-range of one another.Devices cooperate when in radio-range of one another.

http://dblab.usc.eduhttp://dblab.usc.edu

Home-to-Home Online (H2O) devices collaborate to Home-to-Home Online (H2O) devices collaborate to deliver continuous media:deliver continuous media:

A H2O device is a wireless device with a powerful A H2O device is a wireless device with a powerful processor and abundant amount of storage.processor and abundant amount of storage.

Example: H2O CloudsExample: H2O Clouds

S. Ghandeharizadeh, H2O Clouds: Issues, Challenges and Solutions, in IEEE Pacific-Rim S. Ghandeharizadeh, H2O Clouds: Issues, Challenges and Solutions, in IEEE Pacific-Rim Conference on Multimedia, 2003.Conference on Multimedia, 2003.

http://dblab.usc.eduhttp://dblab.usc.edu

Cost of a clipCost of a clip Euclidean distance between the nearest Euclidean distance between the nearest

replica and the requesting node. replica and the requesting node. In a grid, the Euclidean distance between In a grid, the Euclidean distance between

two nodes is defined as:two nodes is defined as:

http://dblab.usc.eduhttp://dblab.usc.edu

Euclidean distanceEuclidean distance Euclidean distance Euclidean distance

between the orange between the orange and green nodes is:and green nodes is: 2222 + 3 + 32 2 = 4 + 9 = 13= 4 + 9 = 13 SQRT (13) = 3.61SQRT (13) = 3.61

http://dblab.usc.eduhttp://dblab.usc.edu

Euclidean distance: 1 replicaEuclidean distance: 1 replica Maximum Maximum

Euclidean distance Euclidean distance to the green node to the green node containing a containing a replica of the replica of the required data item required data item is:is: 2222 + 2 + 22 2 = 4 + 4 = 8= 4 + 4 = 8 SQRT (8) = 2.83SQRT (8) = 2.83

http://dblab.usc.eduhttp://dblab.usc.edu

Euclidean distance: 4 replicasEuclidean distance: 4 replicas Maximum Maximum

Euclidean distance Euclidean distance to the green node to the green node containing a containing a replica of the replica of the required data item required data item is:is: 1122 + 1 + 12 2 = 1 + 1 = 2= 1 + 1 = 2 SQRT (2) = 1.41SQRT (2) = 1.41

Four fold increase Four fold increase in number of in number of replicas provided replicas provided only a factor of 2 only a factor of 2 improvement in improvement in Euclidean Euclidean distance.distance.

http://dblab.usc.eduhttp://dblab.usc.edu

Main assumptionMain assumption1.1. Claim: In a 2-D mesh network, the Euclidean Claim: In a 2-D mesh network, the Euclidean

distance between the nearest replica and the distance between the nearest replica and the requesting node is:requesting node is:

ddii is density of replicas for the i-th object. is density of replicas for the i-th object. Can you provide a proof for this? (Potential Can you provide a proof for this? (Potential

project for the theoreticians in the class.)project for the theoreticians in the class.)

2.2. Ignores issues risen from the boundary by Ignores issues risen from the boundary by assuming an unbounded network.assuming an unbounded network.

http://dblab.usc.eduhttp://dblab.usc.edu

GreedyDual AlgorithmGreedyDual AlgorithmL = 0L = 0

For each request for object i doFor each request for object i do

If i is in the local cache then H(i) = L + C(i)If i is in the local cache then H(i) = L + C(i)

Else while there is no space for i doElse while there is no space for i do

L = min{ H(j) such that j is in cache }L = min{ H(j) such that j is in cache }

Evict j which satisfies H(j) = LEvict j which satisfies H(j) = L

Retrieve and store iRetrieve and store i

H(i) = L + C(j)H(i) = L + C(j)

When object size is variable then replace C(i) When object size is variable then replace C(i) with C(i)/object-size.with C(i)/object-size.

http://dblab.usc.eduhttp://dblab.usc.edu

GreedyDual AlgorithmGreedyDual Algorithm Consider a node with storage capacity of 3 Consider a node with storage capacity of 3

objects. Its storage is occupied with objects objects. Its storage is occupied with objects A, B, and D A, B, and D

Cost of objects is as follows:Cost of objects is as follows: C(A) = 12, C(B)=7, C(E)= 5, C(D)=1C(A) = 12, C(B)=7, C(E)= 5, C(D)=1

Reference for object E causes the algorithm Reference for object E causes the algorithm to evict D because it has the lowest cost, L is to evict D because it has the lowest cost, L is set to 1, H(E)=6.set to 1, H(E)=6.

If the next reference is again for E then If the next reference is again for E then H(E)=?H(E)=?

http://dblab.usc.eduhttp://dblab.usc.edu

GreedyDual AlgorithmGreedyDual Algorithm Consider a node with storage capacity of 3 Consider a node with storage capacity of 3

objects. Its storage is occupied with objects objects. Its storage is occupied with objects A, B, and D A, B, and D

Cost of objects is as follows:Cost of objects is as follows: C(A) = 12, C(B)=7, C(E)= 5, C(D)=1C(A) = 12, C(B)=7, C(E)= 5, C(D)=1

Reference for object E causes the algorithm Reference for object E causes the algorithm to evict D because it has the lowest cost, L is to evict D because it has the lowest cost, L is set to 1, H(E)=6.set to 1, H(E)=6.

If the next reference is again for E then If the next reference is again for E then H(E)=7H(E)=7

What object is evicted if D is referenced? What object is evicted if D is referenced?

http://dblab.usc.eduhttp://dblab.usc.edu

GreedyDual AlgorithmGreedyDual Algorithm Consider a node with storage capacity of 3 Consider a node with storage capacity of 3

objects. Its storage is occupied with objects objects. Its storage is occupied with objects A, B, and D A, B, and D

Cost of objects is as follows:Cost of objects is as follows: C(A) = 12, C(B)=7, C(E)= 5, C(D)=1C(A) = 12, C(B)=7, C(E)= 5, C(D)=1

Reference for object E causes the algorithm Reference for object E causes the algorithm to evict D because it has the lowest cost, L is to evict D because it has the lowest cost, L is set to 1, H(E)=6.set to 1, H(E)=6.

If the next reference is again for E then If the next reference is again for E then H(E)=12H(E)=12

What object is evicted if D is referenced? B, What object is evicted if D is referenced? B, L is now 7, H(D) = 8L is now 7, H(D) = 8

http://dblab.usc.eduhttp://dblab.usc.edu

Modified GreedyDualModified GreedyDual Change the algorithm to not increase H(i) Change the algorithm to not increase H(i)

when it is referenced frequently:when it is referenced frequently:

L = 0L = 0

For each request for object i doFor each request for object i do

If i is in the local cache then do nothingIf i is in the local cache then do nothing

Else while there is no space for i doElse while there is no space for i do

L = min{ H(j) such that j is in cache }L = min{ H(j) such that j is in cache }

Evict j which satisfies H(j) = LEvict j which satisfies H(j) = L

Retrieve and store iRetrieve and store i

H(i) = L + C(j)H(i) = L + C(j)

http://dblab.usc.eduhttp://dblab.usc.edu

EvaluationEvaluation Assumes a fixed square area.Assumes a fixed square area. 40,000 nodes in the mesh network, 40,000 nodes in the mesh network,

distributed randomly.distributed randomly. A node may store 25 objects.A node may store 25 objects. Number of objects is 10,000 with Zipf Number of objects is 10,000 with Zipf

distribution of access.distribution of access. Dijkstra’s shortest path algorithmDijkstra’s shortest path algorithm

Link cost = physical distanceLink cost = physical distance

On average a node has more than 5 directly On average a node has more than 5 directly connected neighbors.connected neighbors.

Normalized access cost = cost / diagonal of Normalized access cost = cost / diagonal of the square area.the square area.

http://dblab.usc.eduhttp://dblab.usc.edu

Performance ResultsPerformance Results

More skewedMore skewed

http://dblab.usc.eduhttp://dblab.usc.edu

Performance ResultsPerformance Results

http://dblab.usc.eduhttp://dblab.usc.edu

Performance resultsPerformance results What other number might be reported?What other number might be reported?

http://dblab.usc.eduhttp://dblab.usc.edu

Performance resultsPerformance results What other number might be reported?What other number might be reported?

Number of replacements performed by each Number of replacements performed by each node. node.

Percentage hit ratio for each node: average & Percentage hit ratio for each node: average & standard deviation.standard deviation. Cold start, and finishCold start, and finish

Was a node thrashing?Was a node thrashing? Is it possible for a node to thrash?Is it possible for a node to thrash?

http://dblab.usc.eduhttp://dblab.usc.edu

Possible project 1Possible project 1 A literature survey of dynamic replacement A literature survey of dynamic replacement

policies for H2O networks. It must include a policies for H2O networks. It must include a comparison with: comparison with: Memory management techniques for an Memory management techniques for an

audio/video-on-demand server.audio/video-on-demand server. Cache servers designed for the Internet.Cache servers designed for the Internet. Cache servers designed for environments such Cache servers designed for environments such

as Akamai.as Akamai.

http://dblab.usc.eduhttp://dblab.usc.edu

Possible project 2Possible project 2 Compare GreedyDual with a technique such Compare GreedyDual with a technique such

as LRU:as LRU: How much improvement does GreedyDual How much improvement does GreedyDual

provide?provide? How does this improvement change as a How does this improvement change as a

function of storage capacity of a node?function of storage capacity of a node? Does one technique exhibit a thrashing Does one technique exhibit a thrashing

behavior?behavior?

How would you set the experiment up?How would you set the experiment up? What would be the important parameters to What would be the important parameters to

quantify and analyze?quantify and analyze? Remember the 10 Commandments when Remember the 10 Commandments when

attempting this project.attempting this project.

http://dblab.usc.eduhttp://dblab.usc.edu

Class Collaborative ProjectClass Collaborative Project Theme: What if cars became driver-less by Theme: What if cars became driver-less by

the year 2100?the year 2100? What would information services would be What would information services would be

available in the car?available in the car? How would the car be used as an extension of a How would the car be used as an extension of a

home/office?home/office?

Deliverable:Deliverable: A script that serves as the basis of a video-clip A script that serves as the basis of a video-clip

(similar to AutoMatas) describing the future (similar to AutoMatas) describing the future information service in cars that drive themselves. information service in cars that drive themselves.

What signs are there that cars will be able to What signs are there that cars will be able to drive themselves?drive themselves?


Recommended