• based on book: "Protocols and Architecture for Wireless Sensor Networks" by Holger and Willig.
  • Egenstudie fra Espen Steine, Thomas Fagerland Wiig og Øystein Taskjelle, H09

This course is planned as a regular course. If you are interested, take contact with Knut Øvsthus, Øivind Kure or Josef Noll

  • Tar utgangspunkt i 2 scenario (per 23. november):
Scenario 1:
  • Areal: 1 x 1 km
  • Tetthet: 5-10 noder per cluster (20 cluster totalt)
  • Dynamisk nettverkstopologi - nodene beveger seg i forhold til hverandre. Lavt energiforbruk og bruker batteri. Ett batteri per levetid.
  • Sink, er mobil, skal kunne hente ut data fra hvilken som helst node i nettverket. Lavt energiforbruk. Bruker batteri. Er ikke fast deltaker i nettverket.
  • Levetid: 1 år
  • QoS: Emergency data skal bli prioritert foran logg data
Scenario 2:
  • Areal: 5 x 5 km
  • Tetthet: 1 km grid
  • Statisk nettverkstopologi - Batteri byttes ved behov.
  • Sink er også statisk. Fast deltager i nettverket. Henter ut data hvert 5 min. (f.eks. værdata). Har fast strømtilkobling.
  • Levetid: 1-2 år
  • QoS: Ingen

Fokus:

  • Definitions: copy to wiki
  • wake-up radio (chapter 4)
  • Reduction of power consumption
Add Comment 
Sign as Author 

Josef Noll, UniK? — 27 November 2009, 15:02

Scenario 1: dynamikk, engery, MAC protokoll,

Lecture Notes

(Discussion on physical layer security (16. Nov 2009) - Attach:20091116PhysLayerSecurity.pdfΔ), Attach:LaserModulations.pdfΔ )

Chapter 1 & 2

Definition

  • Actuator: En utløser enhet (f.eks. bryter)
  • Node: En enhet i nettverket som minimum har en tranceiver, controller, minne og strømkilde. Er et generelt begrep på noe som kan sende og motta.
  • Relay: En node som videresender
  • Sensor: Kan gjøre en måling (f.eks. basert på en request), og omgjør målingen til et elektrisk datasignal.
  • Tag: En enhet som kun sender data.
  • Device: Elektrisk innretning som er laget for et bestemt formål.
  • Sensor node = sensor + node: En node som minimum inneholder en sensor enhet (se definisjon av node og sensor).

Discussions

  • effektforbruk, kom
  • 1 mW utsendt effekt krever 174 mW energi (tab 2.4 på side 73)

Task

  • list typical numbers for power consumption in operation (idle, sleep, Tx, Rx):
    • Tranceiver, receiver og idle state bruker omtrent like mye strøm på lav sendeeffekt.

Formelen som er brukt i de to første parameterne er:


Formel for amplifier power usage (Ref. s 43).

  • Typiske verdier for mikrokontrollere: Intel StrongARM - normal/operational mode 400 mW, i idle mode 100 mW og i sleep mode 50 uW. Texas Instrument 430 - Operational mode 1.2 mW, dypeste sleep mode 0.3 uW, nest dypeste sleep mode 6 uW. (Ref s. 38).
  • Eksempler på strømkilder kan sees i tabellen over.
  • Requirements for operating system and relation to power consumption

Chapter 3

Discussion

  • sender mindre energi
  • simulering (sensor nett)

There are four optimization goals in WSN:

  • QoS,
  • energy efficiency,
  • scalability (autoconfiguration, optimisation) and
  • robustness

Open issues

  • Scenario a) industrial
  • Scenario b) mobile phone as gateway to sensors
  • Scenario c) ...?

Simulation

  • 4 sensors in a row
  • communication from 1-2-3-4 versus 1-2-4 versus 1-4
  • power consumption and radio consumption as a function of distance between sensors
  • Simulation of network design: infrastructure versus ad-hoc

Totalt energiforbruk i WSN med forskjellig antall hop og antall noder, med distanse 100m. Tar utgangspunkt i energiforbruket til tranceiveren uAMPS-1. Bruker følgende foresetninger:

  • 16 PSK 3/4 koderate
  • Avstand 100 meter
  • BER krav på 10^-6
  • Rate på 1 Mbps
  • Pathloss exponent = 3.5 (rullestein -> måler havnivå på ei strand på Hawaii) s98.
  • Pathloss formel s. 93
  • Frekvens = 2.4 GHz
  • N_0 = -180 dB
Tre scenario:
  1. 4 noder, 3 hopp. Resultat: 5.64 J
  2. 5 noder, 4 hopp. Resultat: 4.93 J
  3. 6 noder, 5 hopp. Resultat: 5.15 J

Resultatene viser at det mest energieffektive oppsettet med dette scenarioet er 5 noder - altså 4 hopp. For den interesserte leser kan filen åpnes i SpeQ Mathematics (freeprog). Attach:SimulationEnergyConsumption.txtΔ

Chapter 4

Chapter 5

Chapter 6

Chapter 7

Scenario 1:

  • The approximate size is known on beforehand, and due to the limited number of nodes it is possible to set UID for all nodes.
  • Not quite sure what data we want to get from this kind of network? If we are supposed to receive sort an averaged value from several nodes, the network unique ID is not necessary, but only locally unique IDs.

Scenario 2:

  • This is also a limited sized, in terms of number of nodes, network. Thus, we can preassign network wide unique addresses.
  • Geographic addressing should be interesting here, since the nodes are deployed over an known area and it may be of interest to get a value of a part of the area.

Chapter 8

Chapter 9

Scenario 1:

  • Positioning in dynamic mobile networks is very difficult to achieve, and the proposals available results in a substantial overhead.
  • Static anchors are required to give an absolute position (coordinates according to a reference frame)
  • In this scenario, nodes may not always be in range of 3 or more anchors, and an adaptive schemes that let nodes act as an 'anchor' must be available.
  • Such adaptive schemes results in a high degree of computationally effort, which increase the battery consumption.

Scenario 2:

  • In a static scenario like this one, positioning is much easier, since most nodes already know their position.
  • If the nodes are unaware of their own position, static anchors positioned at the network border could be used as in scenario 1, to compute each nodes position.
  • Static scenarios require less energy consumption
  • Easier to achieve high accuracy

Chapter 10

CORRECTION! At part 3.2 beneath the numbered point list, I wrote : "The black nodes and the edges between them is not forming a dominating set". It should be "The black nodes and the edges between them is now forming a dominating set".

Scenario 1

  • There are already clusters formed naturally, so this would be a good solution.
  • It is important that there exists good methods for merging networks together, if some clusters loose connectivity with other clusters and later rejoin.
  • Passive clustering is an energy efficient method, and due to the dynamics of the nodes there will be a lot of updates regarding neighbors-connectivity which results in routing updates etc. By combining several message in each broadcasted or flooded message, it will save energy rather than send individual messages for each protocol or mechanism.
  • Adpative Self-Configuring sEsnor Network's Topologies might fit this scenario well. Here, the source nodes are able to send HELP-message to improve transmission quality by adding more nodes in the transmission.

Scenario 2

  • Power control that limits the range to the nearest nodes, a k-NEIGH protocol that limits the number of neighbors to a node.
  • Use multihop towards the sink.
  • Having a backbone (rotating the roles) like a dominating set would ease the data-aggregation in the network which I presume is of current interest in this scenario.

Chapter 11

Scenario 1:

  • Because of the dynamic behaviour of this network, nodes need to update its routes more often
  • Nodes need to learn their neighboring nodes before deciding where to send
  • More energy consuming
  • A routing protocol that maximizes the network lifetime should be implementet, altough this protocol may not always give the shortest paths. Since the sink only reads data now and then from the network, a higher latency to save energy is acceptable.

Scenario 2:

  • The nodes are static
  • Minimal changes in topology
  • Its easier for the nodes to determine the best path to the destination, since the topology is static, and the nodes only need to learn about its neighbors one time. After that; only update when/if there are changes in the topology.
  • The sink is static in this scenario, and node batteries are changed when needed. A routing protocol that are able to give more frequent updates and shorter paths between nodes should then be chosen, since battery capacity is not that important anymore. A mesh routing protocol like Core-Assisted Mesh Protocol (CAMP) could then be used.

Chapter 12

Chapter 13

Scenario 1

  • This is a highly dynamic network.
    • Coverage and deployment - nodes move all the time, and the connectivity might be lost between the clusers.
    • Reliable data transport. Both single data and block data is of current interest. Let us say that the nodes have logged data over a period and a user (sink) wants to receive this block data.
    • Congestion control and rate control. Congestion control is needed, but congestion might be limitid due to clustering and data aggregation.

Scenario 2

  • This is a static network
    • Coverage - the average of the data from all nodes gives a kind of coverage of the area. On one hand, the nodes needs to "touch" the environment (rainfall, temperature, wind), and the only coverage is at the exact point of the node, but on the other hand we see that it is more interesting getting an average value over an area which would give a more precise understanding of how's the wheater like.
    • Deployment - in grids, but the important part here is the place the nodes so that they actually can sense what they are supposed the sense (e.g. do not put the rain sensor under a tree).
    • The data coming from the network are mostly single packets (due to aggregation) and no block handling algorithm is needed unless there are software updates to the nodes. This might be needed, and then it would be a good choice with PSFQ. Hard to say whether RMST is better or worse based on the information from the book.
    • Congestions should not appear due to the limited data report rate, but at a scenario where a software update is pushed out to the nodes at the same time as a report is coming in, it might be congestions.