Download presentation

Presentation is loading. Please wait.

Published byNathan Hunt Modified over 3 years ago

1
PROGETTO PATTERN Attività di Ricerca del gruppo di Catania nel I Anno Energy efficient solutions for connectivity and data delivery in self-organizing networks University of Catania Dipartimento di Ingegneria Informatica e delle Telecomunicazioni © Laura Galluccio 04

2
Progetto PATTERN Firenze 12/10/042 Outline Energy efficiency and Timeliness of Neighbor Discovery in Self-Organizing Networks Integrated MAC/Routing protocol for Geographical Forwarding in Self-Organizing Networks

3
Progetto PATTERN Firenze 12/10/043 Energy efficiency and Timeliness of Neighbor Discovery in Self- Organizing Networks

4
Progetto PATTERN Firenze 12/10/044 The Problem Ad hoc and sensor networks: nodes mobility and failures lead to very dynamic topology If nodes are able to discover very rapidly….. ……Self-organization can be achieved….. ….but Higher responsiveness is paid in terms of energy consumption!

5
Progetto PATTERN Firenze 12/10/045 Scenario N2N2 N1N1 R d N1,N2

6
Progetto PATTERN Firenze 12/10/046 Scenario N1N1 N2N2

7
Progetto PATTERN Firenze 12/10/047 Scenario A and B enter each others coverage range they become neighbors N1N1 N2N2 t=0

8
Progetto PATTERN Firenze 12/10/048 Scenario A and B exit each others coverage range they are no longer neighbors N1N1 N2N2 t=

9
Progetto PATTERN Firenze 12/10/049 Scenario N1N1 N2N2

10
Progetto PATTERN Firenze 12/10/0410 Neighborhood Time Neighborhood time (): time interval during which a pair of nodes are neighbors. Discovery time (T): time interval necessary for the discovery between two nodes. Transmit time (T T ): time interval necessary for the exchange of the data required by the application. In case of correct functioning: T T -T Probability of discovery success (PDS): probability that the above relationship holds

11
Progetto PATTERN Firenze 12/10/0411 The Approach A Trade-off is needed between increasing the PDS and decreasing the energy cost associated to this process. We derived a Markov analytical framework for evaluating the energy cost of the hunting process which is the process of searching for other nodes in the proximity. This framework allows the developer to design the hunting process so as to meet the requirements in terms of the energy consumption

12
Progetto PATTERN Firenze 12/10/0412 If remains constant in direction and magnitude for a sufficient amount of time: and

13
Progetto PATTERN Firenze 12/10/0413 The Hunting Process To discover neighbors, a node periodically runs a set of procedures which we call hunting process. The states of this process could be: Inquiry: a beacon message is transmitted (I) Inquiry Scan: a node listens for beacon messages (S) Doze: no inquiry or scan procedures are executed (D) If M channels are used for discovery, the hunting process can be described as: state space state of the hunting process transition rate matrix steady state array

14
Progetto PATTERN Firenze 12/10/0414 Neighbor Discovery Process (NDP) Suppose that in t the two nodes N 1 and N 2 become neighbors. NDP is the result of the interactions between the 2 hunting processes

15
Progetto PATTERN Firenze 12/10/0415 Conditions for Discovery 1. 2. 3. 4.

16
Progetto PATTERN Firenze 12/10/0416 Probability of discovery and array of discovery rate overall discovery rate at t

17
Progetto PATTERN Firenze 12/10/0417 Evolution of the process when nodes do not succeed in discovery The probability that the state of the pair of neighbors is given that the 2 nodes have not discovered We derived matrix which is the state transition rate matrix of the process, given that the mobile nodes have not discovered each other. This matrix can be related to Consequently:

18
Progetto PATTERN Firenze 12/10/0418 QoS Parameters Energy Cost Pdf of the discovery time T eigenvalues of eigenvectors of which is the matrix of the transition rates, given that the mobile nodes do not discover each other

19
Progetto PATTERN Firenze 12/10/0419 PDS in data transfer

20
Progetto PATTERN Firenze 12/10/0420 Cycle Time The cycle time can be related to the scan, inquiry and doze probabilities. It can be demonstrated that, once the probabilities are set, does not depend on or or.

21
Progetto PATTERN Firenze 12/10/0421 Cycle Time

22
Progetto PATTERN Firenze 12/10/0422 Case Study We applied the proposed methodology to a relevant case study: A single channel system We derived some design implications on neighbor discovery algorithms and verified that increases as the cost constraint increases and the ratio decreases.

23
Progetto PATTERN Firenze 12/10/0423 Case Study C*=0.3 C*=0.7

24
Progetto PATTERN Firenze 12/10/0424 Case Study

25
Progetto PATTERN Firenze 12/10/0425 Case Study

26
Progetto PATTERN Firenze 12/10/0426 Conclusions 1 1.Nodes in self-organizing ad hoc and sensor networks execute the hunting process procedures 2.These procedures imply high energy consumption if a timely discovery occurs 3.A trade-off is needed 4.An analytical framework for evaluation of the maximum probability of discovery success, given some energy constraints, was introduced 5.This framework can be used for designing appropriately the discovery process in a mobile network so as to satisfy the application requirements

27
Progetto PATTERN Firenze 12/10/0427 Integrated MAC/Routing protocol for Geographical Forwarding in Self- Organizing Networks

28
Progetto PATTERN Firenze 12/10/0428 The Problem In commercial sensor devices different TX power levels are available This feature can be exploited for reducing energy consumption We propose a MAC/ROuting protocol which uses this capability This protocol works with a cross-layer approach, is simple and does not require any location information knowledge Uses competition to select the most efficient next relay

29
Progetto PATTERN Firenze 12/10/0429 Prerequisites Weighted progress factor (WPF) Set of power levels and coverage range R R D

30
Progetto PATTERN Firenze 12/10/0430 Protocol Functioning Routing functionalities Weighted progress factor (WPF) Set of power levels and coverage range R R D 1.To select the next relay node R triggers a competition 2.Be R the winner of the competition in the set S 1 (R) and G RR1 its WPF 3.If R estimates that a higher WPF can be obtained increasing P, a new competition is triggered in the set S 2 (R). 4.The procedure is repeated until no better relay nodes can be found 5.When the best relay is identified the information is transmitted

31
Progetto PATTERN Firenze 12/10/0431 Protocol Functioning MAC functionalities Weighted progress factor (WPF) Set of power levels and coverage range 1.Nodes periodically switch ON and OFF to reduce energy consumption 2.Synchronization is not needed 3.A wake up phase is requires for R identifying the best relay node in S i (R) 4.To this purpose R transmits several short WAKE-UP messages for a T Cycle 5.Then R sends a Go MESSAGE which triggers competition among nodes in S i (R) 6.A node in Si(R) hearing the WAKE_UP messages calculates its WPF and stays awake waiting for the GO MESSAGE 7.Then upon hearing the GO MESSAGE a node sends randomly back to R its WPF so that R performs the choice of the best relay in S i (R)

32
Progetto PATTERN Firenze 12/10/0432 Questions Analytical Framework What is the probability that outside the coverage area obtained using, exists at least one node whose WPF is higher than, provided that Once the probability is known, when it is worth enlarging the coverage area? If we stop the competition at time t, what is the probability to choose not the best available relay node in the considered coverage area?

33
Progetto PATTERN Firenze 12/10/0433 Answer to Question 1 Where is the nodes density and is the area where a node B must be located in order to belong to and have a WPF higher than g.

34
Progetto PATTERN Firenze 12/10/0434 Answer to Question 2 Where is the probability density function which can be evaluated using the previous results as

35
Progetto PATTERN Firenze 12/10/0435 Answer to Question 3 This represents the probability that a node having a WPF higher than the best available one exists out of and its CONTROL ACK message arrives later than other nodes messages.

36
Progetto PATTERN Firenze 12/10/0436 Performance evaluation Policy for the TX power level choice Policy 1: TX power levels increase linearly Policy 2: TX power levels give a linear increase in Policy 3: TX power levels give a linear increase in

37
Progetto PATTERN Firenze 12/10/0437 Performance evaluation Policy for the TX power level choice Average TX power Average power consumption Average range

38
Progetto PATTERN Firenze 12/10/0438 Performance evaluation Comparison with other protocols Average power consumption Average number of hops

39
Progetto PATTERN Firenze 12/10/0439 Performance evaluation Impact of node density Average power consumption vs. nodes nominal radio area Ninra=NR 2 / 2

40
Progetto PATTERN Firenze 12/10/0440 Performance evaluation Number of packets delivered at destination

41
Progetto PATTERN Firenze 12/10/0441 Conclusions 2 1.MACRO is an integrated MAC/Routing geocast protocol 2.Convenient in case of strict requirements in terms of energy efficiency 3.The increase in end-to-end delay can turn into an advantage in terms of data aggregation 4.MACRO greately extends nodes lifetime

42
Progetto PATTERN Firenze 12/10/0442 Publications L. Galluccio, A. Leonardi, G. Morabito, S. Palazzo: Tradeoff between Energy- Efficiency and Timeliness of Neighbor Discovery in Self-Organizing Ad Hoc and Sensor Networks. Proc. of HICSS 2005, Big Island, Hawaii, January 2005. L. Galluccio, A. Leonardi, S. Palazzo: Design Guidelines for Geocast Protocols in Ad Hoc and Sensor Networks. Proc. of Med-Hoc Net 2004, Bodrum, Turkey, June 2004. L. Galluccio, S. Palazzo: A Taxonomy of Location Management Schemes in Mobile Ad Hoc Networks. To appear in Journal of Communications and Networks. D. Ferrara, L. Galluccio, A. Leonardi, G. Morabito, S. Palazzo: MACRO: An Integrated MAC/Routing Protocol for Geographical Forwarding in Wireless Sensor Networks. Submitted for Publication.

Similar presentations

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google