Postadress: Besöksadress: Telefon:
Box 1026 Gjuterigatan 5 036-10 10 00 (vx) 551 11 Jönköping
Comparison between single mesh network
and cell-based mesh network
Author: Manish Timilsina
THESIS WORK 2011
Electrical Engineering
Postadress: Besöksadress: Telefon:
Box 1026 Gjuterigatan 5 036-10 10 00 (vx) 551 11 Jönköping
Postadress: Besöksadress: Telefon:
Box 1026 Gjuterigatan 5 036-10 10 00 (vx) 551 11 Jönköping
This thesis work is performed at Jönköping University within the subject area of Electrical Engineering. The work is a part of the master’s degree programme with the specialization in Embedded Systems.
The author is responsible for the given opinions, conclusions and results. Examiner: Alf Johansson
Supervisor: Professor Youzhi Xu Scope: 30 Credit Points
Postadress: Besöksadress: Telefon:
Box 1026 Gjuterigatan 5 036-10 10 00 (vx) 551 11 Jönköping
Acknowledgement
Nobody could forget what other did for his or her work. In the same way I cannot forget those people who were always there, supporting suggesting encouraging me to accomplish my project with best result.
I cannot forget my thesis supervisor Professor Youzhi Xu who provided me an opportunity to work with him. I would also like to thank all my friends and seniors for their valuable comments and suggestions throughout the work period. The most important thing in doing the project was the environment that I got. So I cannot forget the respective people who provided me with such environment. I am very thankful to them who always tried keeping the good environment around me throughout the project duration.
i
Abstract
The theme of this thesis is to analyse the performance of a conventional mesh topology in a multipath fading environment and compare it with a newly proposed multiple cell based mesh topology. The communication performance in general is measured by the overall through-put, packets delivery reliability, average message delivery delay and power-consumption.
In this thesis, for simplicity of the calculation the network performance is indirectly measured in-terms of the number of additional routes originally required to connect an isolated or disconnected device, percentage of the devices which have reliable and unreliable route from or to the back bone routers, number of hops from back-bone routers to the nodes and redundant routes which includes the routes inside the particular cell or outside to the other cell.
In this simulation 240 nodes has been used within the area of 120 x 60 m2 which is just in accordance with an average size of industry. Network simulation is broken down into five different scenarios with respect to different number of field devices or nodes and back bone routers along with the presence of obstacles in the area and then analysed respectively. Entire simulation and analytical work have been done on MATLAB.
Major applications of multiple cell based mesh topology can be used within industrial process automation, such as pulp and paper, steel, oil and gas, etc.
ii
Contents
1
Introduction ... 1
1.1 BACKGROUND ... 1
1.2 AIMS AND OBJECTIVES ... 2
1.3 PURPOSE OF DESIGN ... 2 1.4 METHODOLOGY ... 3 1.5 DELIMITATIONS ... 3
2
Theoretical background ... 4
2.1OVERVIEW ... 4 2.1.1 Wireless scenario ... 5 2.1.2 Upper layers ... 5 2.1.3 Network layer ... 52.1.4 802.2 Link logic control (LLC) layer ... 5
2.1.5 MAC layer... 6
2.1.6 PHY layer ... 6
2.2 PHYSICAL LAYER ... 6
2.2.1 Power measurement... 7
2.2.2 Receiver sensitivity ... 8
2.3 NETWORK LAYER:DIJKSTRA’S ALGORITHM ... 8
2.3.1 Algorithm ... 10
2.3.2 An Example of Dijkstra’s algorithm ... 10
3
Technical Overview ... 13
3.1 INTRODUCTION ... 13
3.1.1 Deployment of field devices (nodes) ... 13
3.1.2 A Summary of Nodes Deployment ... 15
3.2 CHANNEL MODEL ... 17
3.3 DEFINITION OF USED PARAMETERS ... 18
3.3.1 Link Quality ... 18
3.3.2 Signal to Noise Ratio (SNR)... 19
3.3.3 Noise Estimation ... 19
3.3.4 Link Cost ... 21
3.3.5 Implementation ... 21
3.3.6 Results and Analysis ... 22
3.3.7 Neighbor table in detail (8 entries) ... 23
3.3.8 Neighbor Table (2 entries) ... 23
3.3.9 Link Cost Analysis ... 23
3.3.10 Deploying Backbone routers ... 23
3.3.11 Routing Algorithm ... 25
3.3.12 Cell Partition ... 25
3.3.13 Partition Technique ... 25
3.3.14 Result of partition ... 25
4
Scenarios and Simulations ... 27
4.1 INTRODUCTION ... 27
4.2 SIMULATION OF 240 NODES ... 27
4.2.1 Distance between the nodes ... 27
4.2.2 Calculate the Rx power for each node ... 27
4.3 SUB-SCENARIOS ... 28
4.3.1 Sub-simulation 1: ... 28
4.3.2 Sub-simulation 2: ... 30
4.3.3 Sub-simulation 3: ... 32
iii
4.4 OBSTACLE BASED SCENARIO ... 35
4.4.1 Obstacle based simulation ... 35
4.5 REDUNDANT PATH SIMULATION ... 36
4.6 SIMULATION FOR INDUSTRIAL SCENARIO ... 37
4.6.1 Simulation methodology ... 37
5
Findings and analysis ... 41
5.1 SINGLE MESH AND MULTIPLE CELL BASED NETWORK ... 41
5.1.1 Reliable and unreliable links ... 41
5.1.2 Obstacle based Reliable and unreliable links ... 43
5.1.3 Sub-simulations for 60,120,160 and 200 nodes ... 44
5.2 SIMULATIONS FOR INDUSTRIAL SCENARIOS ... 46
5.2.1 Industrial Simulation 1 ... 46
5.2.2 Industrial Simulation 2 ... 47
5.3 FAILURE ANALYSIS ... 47
5.3.1 Link Failure ... 48
5.3.2 Node Failure ... 49
5.3.3 Backbone router failure ... 51
6
Conclusion, Applications and Future Work ... 52
6.1 CONCLUSION ... 52
6.2 APPLICATION ... 52
6.3 FUTURE WORK... 53
6.3.1 Establishment of dynamic routing ... 53
6.3.2 Moving backbone routers ... 53
6.3.3 Implementation over real sensor nodes ... 53
References ... 54
Appendices ... 55
APPENDIX A:C++ CODE FOR AUTO GENERATION OF X AND Y VALUES ... 55
APPENDIX B:PACKET ERROR RATE FOR DIFFERENT PACKET LENGTH AND RECEIVE POWER ... 56
APPENDIX C:FLOWCHART OF SIMULATION ... 57
APPENDIX D:NEIGHBOR TABLE ... 58
APPENDIX E:LINK COST ANALYSIS ... 59
APPENDIX F:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER FOR 240 NODES AND 8 BACKBONE ROUTER ... 64
APPENDIX G:ROUTING TABLE ... 66
APPENDIX H:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER WITH OBSTACLE FOR 240 NODES AND 8 BACKBONE ROUTER. ... 72
APPENDIX I:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER FOR SUB SIMULATION 1(30 NODES). ... 74
APPENDIX J:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER WITH 60 NODES AND 2 BACKBONE ROUTER COMBINATION. ... 76
APPENDIX K:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER WITH 120 NODES AND 4 BACKBONE ROUTER COMBINATION. ... 78
APPENDIX L:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER WITH 160 NODES AND 4 BACKBONE ROUTER COMBINATION. ... 80
APPENDIX M:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER WITH 200 NODES AND 8 BACKBONE ROUTER COMBINATION. ... 83
APPENDIX N:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER AND COMBINATION OF BACKBONE ROUTER FOR INDUSTRIAL SIMULATION 1. ... 85
APPENDIX O:STATISTICAL RESULT WITH RESPECT TO EACH BACKBONE ROUTER AND COMBINATION OF BACKBONE ROUTER FOR INDUSTRIAL SIMULATION 2. ... 85
APPENDIX P:STATISTICAL RESULT FOR LINK FAILURE ANALYSIS. ... 86
APPENDIX Q:STATISTICAL RESULT FOR NODE FAILURE ANALYSIS. ... 92
v
List of Figures
FIGURE 1-1: SIMULATION PROCEDURE OF MATLAB ... 2
FIGURE 2-1: ARCHITECTURE OF IEEE 802.15.4 [5] ... 4
FIGURE 2-2: OSI MODEL OF COMPUTER NETWORKING ... 8
FIGURE 3-1: RANDOM DEPLOYMENT OF NODES ... 14
FIGURE 3-2 : VERIFICATION OF RANDOM DISTRIBUTION OF NODE ... 15
FIGURE 3-3 : ASSUMPTIONS ABOUT TOPOLOGY ... 16
FIGURE 3-4 : TOPOLOGY WITH GATEWAY ... 16
FIGURE 3-5 : TOPOLOGY WITH BACKBONE ... 16
FIGURE 3-6 : X NORMAL DISTRIBUTION ... 17
FIGURE 3-7 : NODES A TO NODE B ... 18
FIGURE 3-8 : NODES B TO NODE A ... 18
FIGURE 3-9 : BRE ATTENUATION WITH PR (D) ... 20
FIGURE 3-10 : PER ATTENUATION WITH PR (D) ... 21
FIGURE 3-11: LINK COST ... 21
FIGURE 3-12 : BACKBONE ROUTERS’ POSITION ... 23
FIGURE 4-1: TOP-TO-BOTTOM:TOP: 30 NODES PARTITION WITH BACKBONE ROUTER 1 AND 2 ... 29
FIGURE 4-2: TOP-TO-BOTTOM:TOP: 30 NODES PARTITION WITH BACKBONE ROUTER 5 AND 6 ... 29
FIGURE 4-3 : TOP-TO-BOTTOM: TOP: 60 NODES PARTITION WITH BACKBONE ROUTER 1 AND 2 ... 30
FIGURE 4-4 : 60 NODES PARTITION WITH BACKBONE ROUTER 5 AND 8 ... 31
FIGURE 4-5 : 120 NODES PARTITION WITH BACKBONE ROUTER 1,2,3 AND 4 ... 32
FIGURE 4-6 : 120 NODES PARTITION WITH BACKBONE ROUTER 5, 6, 7 AND 8 ... 33
FIGURE 4-7 : 200 NODES PARTITION WITH BACKBONE ROUTER 1, 2, 3, 4, 5, 6, 7 AND 8 ... 34
FIGURE 4-8 : OBSTACLE IN THE NETWORK ... 35
FIGURE 4-9 : PARTITIONING OF NODES BETWEEN PAN-COORDINATOR AFTER OBSTACLES ... 36
FIGURE 4-10 : SCENARIO 1: RANDOM DEPLOYMENT OF NODES ... 38
FIGURE 4-11 : SPANNING TREE OF CELL 1 AND CELL 2 USING DIFFERENT BACKBONE ROUTER ... 38
FIGURE 4-12 : DEPLOYMENT OF NODES ... 39
FIGURE 4-13 : DEPLOYMENT OF BACKBONE ROUTER ... 39
FIGURE 4-14 : SPANNING TREE OF CELL 1 CELL 2 AND CELL 3 USING DIFFERENT BACKBONE ROUTER . 40 FIGURE 5-1: 1 HOP ROUTES VS MULTIPLE HOP ROUTES ... 41
FIGURE 5-2: ROUTE COST COMPARISION ... 42
FIGURE 5-3: 1 HOP ROUTES VS MULTIPLE HOP ROUTES ... 42
FIGURE 5-4: 1 HOP ROUTES VS MULTIPLE HOP ROUTES ... 43
FIGURE 5-5: ROUTE COST COMPARISION ... 43
FIGURE 5-6: MAXIMUM NUMBER OF HOPS VS AVERAGE HOP ... 43
FIGURE 5-7: SUB-SIMULATION FOR 1HOP ROUTES VS MULTIPLE HOP ROUTES ... 44
FIGURE 5-8: SUB-SIMULATIONS FOR THE HOP ANALYSIS ... 45
FIGURE 5-9: SUB-SIMULATION FOR ROUTER COST ... 45
FIGURE 5-10 : SIMULATION FOR INDUSTRIAL SCENARIO 1 ... 46
FIGURE 5-11: SIMULATION FOR INDUSTRIAL SCENARIO 2 ... 47
FIGURE 5-12 : LINK FAILURE ANALYSIS - MAXIMUM VS. AVERAGE HOP ... 48
FIGURE 5-13 : LINK FAILURE ANALYSIS – 1 HOPS VS. MULTIPLE HOPS ... 48
FIGURE 5-14: LINK FAILURE ANALYSIS-AVERAGE RC COMPARISION ... 49
FIGURE 5-15 : NODE FAILURE- MAXIMUM VS. AVERAGE NUMBER OF HOPS. ... 49
FIGURE 5-16: NODE FAILURE-1 HOP VS. MULTIPLE HOPS ... 50
FIGURE 5-17: NODE FAILURE-ROUTE COST COMPARISION... 50
v
List of Tables
TABLE 3-1 : COMPARISON OF RANDOM DISTRIBUTION OF NODES ... 14
TABLE 3-2 : SAMPLED DISTANCE VALUES W.R.T NODE 0 ... 15
TABLE 3-3: BETA VALUE FOR NODES TO NODES POWER CALCULATION ... 17
TABLE 3-4: CONDITION FOR NON-EXISTENCE OF LQ ... 19
TABLE 3-5 : THE FORMAT OF NEIGHBOR TABLE ... 22
TABLE 3-6 : DISTRIBUTION OF RANDOM VARIABLE X IN SIMULATION ... 22
TABLE 3-7 : BR’S PHYSICAL POSITION ... 24
TABLE 3-8: BETA VALUE FOR NODES FOR POWER CALCULATION FROM AND TO BACKBONE ROUTER 24 TABLE 3-9 : NUMBER OF SENSOR-NODES IN EACH CELL ... 25
TABLE 3-10 : STATISTICAL RESULT OF ROUTER COST IN EACH CELL ... 26
1
1 Introduction
In contrast to the wired networks, wireless networks provide advantages in deployment, cost, size, and distributed intelligence. Wireless technology not only enables users to set up a network quickly, but also enables them to set up a
network where it is inconvenient for human to reach or impossible to wire cables. The care-free feature and convenience of deployment make a wireless network more cost-efficient than a wired network in general. [1]
The information needed by smart environment is provided by wireless sensor networks, a sensor network defined as a cluster of distributed sensors on any large or a small scale to monitor physical or environmental conditions, such as
temperature, sound, vibration, pressure, motion or pollutants. The development of wireless sensor networks was motivated by military applications such as battlefield surveillance. They are now used in many industrial and civilian application areas, including industrial process monitoring and control, machine health monitoring, environment and habitat monitoring, healthcare applications, home automation, and traffic control. [2]
Different wireless protocols were also considered. Applications such as 802.11 Wireless Local Area Network (WLAN) are an inappropriate with a redundant data rate and high power consumption. Bluetooth protocol was introduced in 1994 for a low data rate to reduce cables for computers and mobile devices. The
disadvantage of Bluetooth protocol is the limitation of number of nodes that can be connected simultaneously (1 master & 7 slaves) and the high level of power consumption. A new implementation of wireless sensor network IEEE 802.15.4 and Zigbee introduced in a year 2000 with a main concern of low-data rate control and sensor applications in wireless networks. Zigbee is predicated on IEEE
802.15.4 technological standard for low data rate in the Industrial, Scientific and Medical (ISM) frequency band. Low data rate provided by IEEE 802.15.4, allow communication among devices with consideration to very low power
consumption in use of battery supply. IEEE 802.15.4 devices are appropriate for home environment with a main topic of a low-cost and low data rate. [3]
1.1 Background
The idea of this project is to analyse a multiple cell-based mesh topology which gives the concept of multiple gateways (Backbone routers) on different locations in a network and enable data routing with less number of hops, each backbone router has its own respective nodes and every node is responsible to collect the data from the environment and send it to its respective backbone router and switch to other when required. In-order to make design more real, an obstacle based environment is also considered.
Since it has been decided to be as much as precise in the calculation and make the simulation near to the real-world scenarios; for that different software’s were taken into consideration, and finally it has been decided to be bounded to MATLAB , MATLAB has a very good reputation in defining mathematical models and performing lengthy simulations in a quick time.
2
There are several reasons to work on a MATLAB instead of working with real nodes, since it gives inexpensive, flexible and reconfigurable environment network phenomena, opportunity to study a large scale network and an easier comparison of result across research effort, the simulation procedure normally consists of following steps in figure 1-1.
1.2 Aims and Objectives
Transfer the maximum amount data and make sure a successful reception at the desired node in a multi-path fading environment.
To make sure the maximum number of sensor nodes get connected with the backbone routers with a reliable link or better link cost.
To reduce the number of hops as much as possible between backbone routers and sensor nodes.
Availability of the redundant paths on the failure either of node or path or a backbone router itself.
1.3 Purpose of design
Since there is no any particular topology which can tackle with the congestion, therefore in this project all the work has been done from the scratch by
considering congestion in the network which creates multi-path fading effect.
Definition of Problem Mathematical Model Verification & simulation on MATLAB Finish Simulation False True Literature Review
3
In an obstacle based environment, a mesh topology with a single back-bone router is not a good choice, because the number of hops will get increase, and most of the nodes will never get synchronized with a router because of the transmission delay or obstacle in-between, therefore it has been assumed if multiple back-bone routers introduced around the network, then nodes synchronization and increase in hop counts will be handled more efficiently.
In real-world, the received power at certain distance is a random variable due to multi-path propagation effects, which is also known as multi-path fading effect, the two models in network simulator ,two-ray ground reflection and free space predict the mean received power at a certain distance which is known as a circle of communication between source and target nodes. [4]
Practically it is hard to arrange many sensor nodes, because in industry there are mainly a lot of sensor nodes deployed on many different areas, therefore a simulator has been chosen for an ease of implementation of cell-based mesh network and to create obstacles based scenario a multi-path signal fading effect has been created in Matlab.
1.4 Methodology
The methodology which has been selected for this project work is simulation over the computer , the main reason of selecting this technique is that it gives an easy and quick approach to perform the work and achieved the desired results , the only drawback for this technique is that sometime its unable compute the factors which involve in the real-world scenario. But it is highly cost effective and quick.
1.5 Delimitations
Due to the limited economic resources and tight timing schedule it was very hard to perform the task practically, therefore all the work has been done on Matlab.
4
2 Theoretical background
2.1 Overview
IEEE 802.15.4 defines the physical layer (PHY) and medium access control (MAC) sub-layer specifications for low-data-rate wireless connectivity with fixed, portable, and moving devices very limited battery consumption requirements typically operating in an operating space of 10 m. It is foreseen that, depending on the application, a longer range at a lower data rate may be an acceptable
tradeoff.[5]
Below is the figure 2.1 showing architecture of IEEE 802.15.4.
Wireless Scenerio Definition Upper layer Network layer 802.2 LLC SSCS 802.15.4 MAC 802.15.4 PHY ED CCA LQI Filtrering Multi-Channel CSMA-CA
Beacon and Sync.
Assoc. And Dissoc
Direct/Indirect GTS Tx
Filtrering
Error Modeling
Figure 2-1: Architecture of IEEE 802.15.4 [5]
In-order to understand the above figure 2.1, a brief explanation need to made, but before diving into the explanation it is better break-down the above defined figure into following steps
1) Wireless scenario definition 2) Definition of upper layers 3) Definition of Network layer
5
4) 802.2 Link logic control (LLC) layer 5) SSCS (Service specific convergence layer 6) 802.15.4 MAC layer
7) 802.15.4 PHY layer
2.1.1 Wireless scenario
In this section, a simulating scenario need to be defined which can varied upon different situations, various topologies can also be defined i.e. star, mesh or cluster tree topology. The environment of the topology can either be an obstacle based or obstacle-free. In-order to start the simulation this step is necessary to perform.
2.1.2 Upper layers
This layer usually defined, application, presentation and session layers, which can be defined with respect to the desired protocol, need to be implemented by the programmer for the simulation of topology or testing of any routing protocols.
2.1.3 Network layer
The Network layer is responsible for routing packets delivery including routing through intermediate routers, finding redundant paths, network management, route discovery and maintenance. The routing protocol in this project used by the network layer is dijiktra’s routing algorithm, it is a graph search algorithm that solves the single-source shortest path problem for a graph with nonnegative edge path costs, producing a shortest path tree. This algorithm is often used in routing and as a subroutine in other graph algorithms, the detailed overview about
dijiktra’s algorithm will be discussed later in the following sections.
2.1.4 802.2 Link logic control (LLC) layer
Logic Link Control (LLC) is the IEEE 802.2 LAN protocol that specifies an implementation of the LLC sub-layer of the data link layer, It provides a way for the upper layers to deal with any type of MAC layer. IEEE 802.2 LLC is used to perform the following functions:
Managing the data-link communication
Link Addressing
Defining Service Access Points (SAPs)
Sequencing
In-order to keep the discussion restricted with respect to routing problems in multiple cell based mesh topologies, only network layer is completely focused and explained in the later sections.
6 2.1.5 MAC layer
This clause specifies the MAC sub-layer with respect to the IEEE 802.15.4. The MAC sub-layer handles all access to the physical radio channel and is responsible for the following tasks:
Generating network beacons if the device is a coordinator
Synchronizing to network beacons
Supports association and disassociation between nodes and back-bone router.
Supporting device security
Employing the CSMA-CA mechanism for channel access
Handling and maintaining the GTS mechanism
Providing a reliable link between two peer MAC entities
The MAC sub-layer provides an interface between the SSCS and the PHY. The MAC sub-layer conceptually includes a management entity. This entity provides the service interfaces through which layer management functions may be invoked. It is responsible for maintaining a database of managed objects pertaining to the MAC sub-layer.
2.1.6 PHY layer
The PHY in IEEE 802.15.4 is responsible for the following tasks:
Activation and deactivation of the radio transceiver.
Energy detection (ED) within the current channel.
Link quality indicator (LQI) for received packets.
Clear channel assessment (CCA) for carrier sense multiple access with collision avoidance (CSMA-CA).
Channel frequency selection.
Data transmission and reception.
Further discussion on the physical layer of IEEE-802.15.4 protocol with respect to the project will be discuss in the following sections.
2.2 Physical layer
The standard specifies the following four PHYs:
An 868/915 MHz direct sequence spread spectrum (DSSS) PHY employing binary phase-shift keying (BPSK) modulation
An 868/915 MHz DSSS PHY employing offset quadrature phase-shift keying (O-QPSK) modulation
7
An 868/915 MHz parallel sequence spread spectrum (PSSS) PHY employing BPSK and amplitude shift keying (ASK) modulation
A 2450 MHz DSSS PHY employing O-QPSK modulation
In addition to the 868/915 MHz BPSK PHY, which was originally specified in the 2003 edition of this standard, two optional high-data-rate PHYs are specified for the 868/915 MHz bands, offering a tradeoff between complexity and data rate. Both optional PHYs offer a data rate much higher than that of the 868/915 MHz BPSK PHY, which provides for 20 kb/s in the 868 MHz band and 40 kb/s in the 915 MHz band. The ASK6 PHY offers data rates of 250 kb/s in both the 868 MHz and 915 MHz bands, which is equal to that of the 2.4 GHz band PHY. The O-QPSK PHY, which offers a signaling scheme identical to that of the 2.4 GHz band PHY, offers a data rate in the 915 MHz band equal to that of the 2.4 GHz band PHY and a data rate of 100 kb/s in the 868 MHz band.
2.2.1 Power measurement
All power measurements either transmit or receive, shall be made at an
appropriate transceiver to antenna connector. The measurements shall be made with equipment that is either matched to the impedance of the antenna connector or corrected for any mismatch. For devices without an antenna connector, the measurements shall be interpreted as effective isotropic radiated power (EIRP) (i.e., a 0 dBi gain antenna); and any radiated measurements shall be corrected to compensate for the antenna gain in the implementation.
In term of the mathematical term, the receive power can be measured by channel model equation.
Pr(d)=Pr(d0)-10 βlog( )+X
Where, Pr (d) is the received power to be calculated. d0 is the reference distance, d is the physical distance between transmitter and receiver, Pr (d0) is the received power of the reference distance, β is the signal attenuation coefficient ,and X (in dBm) is a zero-mean normal (or Gaussian) distributed random variable with the standard deviation σ , X~ (0, σ) The log-normal distribution describes the random shadowing effects occurring over quantities of devices’ locations which have the same receiver-transmitter separation, but have different levels of clutter. on the propagation path. The distance-dependent mean is
β and X in this channel model describe the loss path loss for an arbitrary location having a specific Receiver-Transmitter separation. In computer simulation this model can provide received power levels for random locations in communication systems.
(2) (1)
8 2.2.2 Receiver sensitivity
Receiver sensitivity is a threshold input signal power that yields a specified PER (packet error rate).
Some conditions for PER (Packet error rate) are defined below for the condition of -85dbm.
PSDU length = 20 octets.
PER < 1%.
Power measured at antenna terminals.
Interference not present.
2.2.2.1 Packet error rate
Packet error rate is an average fraction of transmitted packets that are not detected correctly.
Or
Packet Error Rate (PER) is a percentage of package that cannot be correctly received. PER can be calculated by the following formula:
Where, ε is BER, and l is the bit-length of the packet.
2.2.2.2 Bit error rate
Bit Error Rate (BER) is the number of bit errors divided by the total number of transferred bits during a studied time interval. BER can be calculated by this formula:
2.3 Network layer: Dijkstra’s Algorithm
The Network Layer is Layer 3 from bottom of the seven-layer OSI model of computer networking. Application layer Presentation layer Session layer Transportation layer Network layer Data link layer Physical layer
Figure 2-2: OSI model of computer networking
(3)
9
The Network Layer is responsible for routing packets delivery including routing through intermediate routers, whereas the Data Link Layer is responsible for Media Access Control, Flow Control and Error Checking.
The Network Layer provides the functional and procedural means of transferring variable length data sequences from a source to a destination host via one or more networks while maintaining the quality of service functions.
Within the service layering semantics of the OSI network architecture the
Network Layer responds to service requests from the Transport Layer and issues service requests to the Data Link Layer.
“Routing is the process of selecting paths in a network along which to send network traffic”. Routing is performed for many kinds of networks, including the telephones, sensor networks, Internet, and transportation networks.
Dijkstra is a shortest path algorithm routing technique finds the lowest-cost path from one node to all others in a graph with weighted edges. It is easiest to think of these as geographical distances, with the vertices being places, such as nodes, as illustrated in figures shown below.
A B C E D 8 2 7 3 10 5
Imagine you are at A, and would like to know the shortest way to get to the
surrounding towns: B, C, D, and E . You would be confronted with problems like: Is it faster to go through D or C to get to E. Is it faster to take a direct route to B, or to take the route that goes through C As long as you knew the distances of paths going directly from one point to another, Dijkstra's algorithm would be able to tell you what the best route for each of the nearby point would be.
10 2.3.1 Algorithm
1. Begin with the source node, and call this the current node. Set its value to 0. Set the value of all other nodes to infinity. Mark all nodes as unvisited.
2. For each unvisited node that is adjacent to the current node, do the following. If the value of the current node plus the value of the edge is less than the value of the adjacent node, change the value of the adjacent node to this value. Otherwise leave the value as is.
3. Set the current node to visited. If there are still some unvisited nodes, set the unvisited node with the smallest value as the new current node, and go to step 2. If there are no unvisited nodes, then we are done.
In other words, we start by figuring out the distance from our hometown to all of the towns we have a direct route to. Then we go through each town, and see if there is a quicker route through it to any of the towns it has a direct route to. If so, we remember this as our current best route. [6]
2.3.2 An Example of Dijkstra’s algorithm
Let A as current node. Give it a value of 0, since it doesn't cost anything to get to it from starting point. Assign everything else a value of infinity, since a way to get to them is unknown. A 0 B ∞ C ∞ E ∞ D ∞ 8 2 7 3 10 5
Next, look at the unvisited points our current node is adjacent to. This means B, D and E. We check whether the value of the connecting edge, plus the value of our current node, is less than the value of the adjacent node, and if so we change the value. In this case, for all three of the adjacent nodes we should be changing the value, since all of the adjacent nodes have the value infinity. We change the value to the value of the current node (zero) plus the value of the connecting edge
11
(10 for E, 5 for D, 8 for B). We now mark A as visited, and set D as our current node since it has the lowest value of all unvisited nodes.[6]
A 0 B 8 C ∞ E 10 D 5 8 2 7 3 10 5
The unvisited nodes adjacent to D, our current node, are C and E. So we want to see if the value of either of those points is less than the value of D plus the value of the connecting edge. The value of D plus the value of the path to E is 5 + 3 = 8. This is less than the current value of E (10), so it is shorter to go through D to get to E. We change the value of E to 8, showing we can get there with a cost of 8. For C, 5 + 7 = 12, which is less than C’s current value of infinity, so we change its value as well. We mark D as visited. There are now two unvisited nodes with the lowest value (both C and E have value 8). We can arbitrarily choose E to be our next current node. However, there are no unvisited nodes adjacent to
Greenville! We can mark it as visited without making any other changes, and make Orangeville our next current node.
A 0 B 8 C 12 E 10 D 5 8 2 7 3 10 5
12
There is only one unvisited node adjacent to B. If we check the values, B plus the connecting road is 8 + 2 = 10, C value is 12, and so we change C value to 10. We mark B as visited, and C is our last unvisited node, so we make it our current node. There are no unvisited nodes adjacent to C, so we're done![6]
A 0 B 8 C 10 E 10 D 5 8 2 7 3 10 5
Here is the same example, with all of the steps laid out in table form:
Current Visited Red Green Blue Yellow Orange Description Red 0 Infinity Infinity Infinity Infinity Initialize Red
as source
Red 0 10 5 8 Infinity Change
values of Green , blue and yellow
Blue Red 0 10 5 8 Infinity Set Red as
visited blue as current
Blue Red 0 8 5 8 12 Change value
for orange
Green Red,blue 0 8 5 8 12 Set blue as
visited blue as current Yellow Red,blue,
Green 0 8 5 8 12 Set green as visited yellow as current Yellow Red,blue,
Green 0 8 5 8 10 Change value for orange Orange Red,blue,
Green,yellow 0 8 5 8 10 Set yellow as visited ,orange as current Red,blue, Green,yellow orange 0 8 5 8 10 Set Orange as visited
13
3 Technical Overview
3.1 Introduction
This chapter will describe Algorithms, calculations and an approach used to solve the obstacle problem in the case scenarios.
The Project work is divided into following steps 1. Deployment of Sensor nodes.
2. Verification of random deployment. 3. Routing algorithm.
4. Addition of backbone routers. 5. Cell-balancing.
6. Deployment of obstacles in the network.
3.1.1 Deployment of field devices (nodes)
This is the most important and the first step of the project work, i.e. to deploy the random topology of 240 nodes in an area of 120x60, the deployment is broken down into 2 steps.
Random deployment of nodes
Verification and Validation of deployment
Random deployment of nodes
A scenario of uniformly randomly distributed 240 nodes has been implemented in the project at this very first stage with unique numbers of X and Y co-ordinates for each node, but since it is difficult to generate unique number of co-ordinate values randomly, therefore a small code has been written in C++ for the auto generation of X and Y values, the code is shown below in Appendix A.
Verification and Validation of deployment
The nodes are randomly deployed in area of 120 m x 60 m showed in figure 3-1 the random deployment of nodes of X-Y plane.
14
Figure 3-1 shows the random deployment of the nodes, this figure has been used to calculate either the nodes are randomly deployed or not which has been calculated manually by using equation 5; entire plane has been broken down into small grids the size of the grid is 1x1 m2. Percentage of grids with K devices in each grid is:
After calculating the values manually the values were verified by using binominal distribution equation 6:
( ) Here, n: number of nodes = 240
a: 1/ (120x60) = 1/7200 = 0.000138 K=0, 1, 2, 3, 4……..n
After putting the above defined values in equation 6 and obtain the values from equation 5, we get the following table.
Node Number X-Coordinate Y-Coordinate Distance
0 2.9 32.6 0 1 51.1 30.6 48.24 2 74.6 46.4 73.01 3 1.8 21.5 11.15 4 89.4 54 89.10 5 49.7 7.2 53.24 6 107 8.6 106.83
Table 3-1 : Comparison of random distribution of nodes
(5)
(6) Figure 3-1: Random Deployment of Nodes
15
In order to save the time and gets the results more accurately, the random
distribution probability function has been implemented in MATLAB as shown in figure 3-2.
In the previous steps the nodes are randomly deployed and verified, but with unknown distance information, in-order to calculate the distance of each node with every other node, below is the equation 7 which has been used.
Here, D = Physical distance X and Y = Co-ordinates of nodes.
Table 3-2 shows the some sample values of physical distance of nodes w.r.t node 0 (taken as first node here) , similar process has been used for other nodes in the network.
Node
Number X-Coordinate Y-Coordinate Distance
0 2.9 32.6 0 1 51.1 30.6 48.24 2 74.6 46.4 73.01 3 1.8 21.5 11.15 4 89.4 54 89.10 5 49.7 7.2 53.24 6 107 8.6 106.83
Table 3-2 : Sampled distance values w.r.t node 0
3.1.2 A Summary of Nodes Deployment
This section will contain the flow chart review of all the work which has been done in the project at the initial steps, the flow chart is shown below.
y=1:5
y=binopdf(0:4,240,0.000139)*100; y'
(7) Figure 3-2 : verification of random distribution of node
16
In industrial networks topology, normally the manger and the mesh-networks are connected by the gateways, like the following picture illustrated.
One of the major disadvantages of this topology is the gateway capacity. The average rate in industrial is 1P/sec, and the size of mesh-network is 30-40 devices, and it means the gateway capacity can limit the network capacity with this
topology.
The solution to solve the gateway capacity is to use multiple backbone routers and partition devices into cells.
Figure 3-4 : Topology with gateway Figure 3-3 : Assumptions about topology
Make Code on C++ for 240 nodes
Deploy on 120 X 60 plane
Verify results
Use the constelation diagram
Verifed
Not verified
Find distance between all nodes
END OF STEP 1
17
In this topology, backbones are added into the mesh-networks, this gives an advantage of reducing hop counts as well, and with this approach network capacity will be possibly increased.
3.2 Channel model
The channel model is the log-normal model with shadowing effects, the formula of this model is shown in equation 1 section 2.2.1. In this project study, the standard deviation σ is set to 10 dBm, then X~ N (0, 10). Theoretically, if , then probability density function f(x)
f(x) =
√
It can be described as the following graph
Figure 3-6 : X normal distribution
The theoretical probability is the reference to the X in computer simulation. The attenuation coefficient β is set to every node randomly in condition that either the transmission is line of sight (LOS) or non-LOS:
ß Value Percentage of node Maximum Range
2 (LOS) 15 % 100 m
3 (LOS) 15 % 100 m
4 (N-LOS) 23.4 % 60 m 5 (N-LOS) 23.3 % 25 m 6 (N-LOS) 23.3 % 15 m
Table 3-3: Beta value for Nodes to Nodes power calculation
18
3.3 Definition of used Parameters
The deployment of 240 nodes has been mentioned in Introduction section. In general, 240 nodes (set to full-functioned devices) are distributed randomly in area of (x-coordinate y-coordinate). For the performance of the
network capacity in this scenario, the neighbor table of each node is required. Several parameters in wireless communication can be considered as the significant key to describe a specific neighbor table. These parameters are Link Quality, Signal to Noise Ratio (SNR), Bit Error Rate (BER), Packet Error Rate (PER), and Link Cost.
3.3.1 Link Quality
If node A is connected to node B, and node A is the transmitter; node B is the receiver.
The link quality (LQ) for the transmission from A to B can be defined as the received power (in dBm) from A to B. According to Eq.1, Pr (d) A->B= Pr (d) = Pr (d0)-10βlog ( ) + X , where dAB is the physical distance between A and B, so
Due to shadowing deviation, when LQA->B is calculated by Eq.1 directly, to calculate LQB->A, a Gaussian random process can be used to consider the shadowing deviation in transmission between B and A, and this Gaussian random process can be described as adding a uniform random distribution
f(y) = , -4<=y<=4 dB
(11)
(9)
Figure 3-7 : Nodes A to Node B
Figure 3-8 : Nodes B to Node A
(9)
19
According to both Eq.8 and Eq.9, the link quality is determined by the attenuation coefficient β, the physical distance d, and also the sensitivity of received power, link cost of many nodes can be defined as non-existence (disconnection) under the condition of receive power less than -97dBm or beyond of shown in the table 3-4 below:
3.3.2 Signal to Noise Ratio (SNR)
Signal-to-noise ratio is a measure used to quantify how much a signal has been corrupted by noise. It is defined as the ratio of signal power to the noise power corrupting the signal.
If the unit of both signal and noise is in dBm, then
3.3.3 Noise Estimation
Noise needs to be estimated to calculate the PER and BER of nodes in networks. In common, there are two approaches to estimate noise:
First approach:
According to Eq.3 and Eq.4:
√
In this case, Physical Service Data Unit (PSDU), which is defined by the IEEE 802.15.4 Standard, is used to estimate the Noise power.
From IEEE 802.15.4 standard, if
PSDU (Physical Service Data Unit), l=160 bits and Packet Error Rate (PER) = 1% then
Receive signal power (S) = -85 dBm but ß Value Physical Distance
2 100 m
3 100 m
4 60 m
5 25 m
6 15 m
Table 3-4: Condition for non-existence of LQ
(12)
20
The measured sensitivity using CC2420 mote is in between -91dBm ~ -94 dBm so it has been decided to use receive signal power as -94 dBm in this thesis.
Using the above parameters and substituting the values in the equation 13 we get Noise, N1 = 4.432099 10-11 mW = -103.53 dBm
Second approach:
According to the noise power equation
N2 = KTB Here,
Boltzmann’s Constant, K = Room Temperature in Kelvin, T = Bandwidth, B = 5 MHz
Due to DSSS modulation technique, the process gain is 8 (i.e. 32chips/4bits), then noise power in this approach
Noise, N2 = 2.57025 10-12 mW = -115.9 dBm.
N1 is set to use in this project, meanwhile N2 can be considered as a reference. Science noise power is fixed to N1, using Eq.3 and the relationship between received power and BER can be found as the following graph
Also, using Eq.4, the relationship between received power and PER can be found, the length of packet data varies from 20 bytes to 120bytes. Detailed table is given in appendix B.
(14)
21 3.3.4 Link Cost
Link cost is defined as the minimum number of transmissions for successful packet data. If node A and node B is connected, and node A is the transmitter, node B is the receiver.
Rate of transmitted Data A->B and the Packet Error Rate of Acknowledgement B->A. Here, the length of both transmitted data and ACK is fixed to be 60bytes (480bits) and 26bytes (208bits) respectively.
Then Link Cost from A to B is
3.3.5 Implementation
In this project, the neighbor table of each node is required to record for improving the routing algorithm. The entries of neighbor table in this project contain:
1) The number of nodes connected to the current node (node #); 2) Link cost
(15) Figure 3-10 : PER Attenuation with Pr (d)
22
Node # Link Cost
J 2.31
R 1.03
… …
Table 3-5 : The format of neighbor table
To record the neighbor table of each node, the equations and methods in previous section are used. The simulation is programming in Matlab to generate and
analyze the neighbor table. The flow-chart in appendix C explains how to make the neighbor table of each node in programming.
In the phase of generating the value of β, if there is a connection between node i and node j, and node i is considered as a transmitter first, and when node j is considered as transmitter
.
In the phase of making link cost symmetrical, if node i and node j are connected to each other, then
3.3.6 Results and Analysis
Random variable X
In this table, the number of node is randomly chosen from 1 to 240. The table is illustrated the compassion between statistical result and theoretical result.
Node # Range of x Node 137 Node 68 Node 214 Theoretical (Eq.2) ( 0,10 ) 32.08% 39.17% 36.25% 34.13% ( -10,0 ) 34.58% 35.42% 32.50% 34.13% ( 10,20 ) 13.33% 11.25% 12.50% 13.59% (-20,-10) 16.67% 7.92% 13.75% 13.59% Table 3-6 : Distribution of Random Variable X in simulation
(16)
23 3.3.7 Neighbor table in detail (8 entries)
The maximum number of neighbors in the neighbor table is 42; the minimum number of neighbors in the neighbor table is 11.The neighbor table for Node 156 is only one example of the whole 240 nodes neighbor tables shown in appendix D Table 1.
3.3.8 Neighbor Table (2 entries)
The neighbor table with symmetrical link cost contains two entries as the format of the neighbor table shown in appendix D Table 2.
3.3.9 Link Cost Analysis
In this section the link cost of nodes has been measured for the analysis of possible neighbors available to each node, but with a certain conditions, i.e. the nodes connected with a link cost in between 1.0 to 2.0 are considered and are shown in the table at Appendix E.
3.3.10 Deploying Backbone routers
The deployment of sensors in WSN is the same one shown in Fig.2, which is 240 nodes in an area with fixed x and y coordinates. In the given position (Table 3-7), add each of the eight nodes to be the backbone router individually as the following graph.
24 Backbone Router # X-coordinate Y-coordinate 1 60 0 2 60 60 3 0 30 4 120 30 5 20 0 6 20 60 7 100 0 8 100 60
Table 3-7 : BR’s Physical Position
Based on the previous Equations (from eq.1 to eq.17), the connection between backbone routers and 240 nodes are simulated, but in this simulation, the signal attenuation coefficient β is set according to the following condition:
Also eq.7 changes to be
√
To calculate the physical distances between every sensor node and the backbone routers.
In case, β is randomly set to 2, 3, 4, 5, and 6 with probability 20% respectively for all nodes in WSN. When adding the first backbone router into WSN, node 1 is the name of the backbone router, and from node 9 to node 248 are the previous nodes from 1 to 240.
The link quality of node i ( ) to Backbone Router j ( ) is calculated by using equation 9, 10 and 11.
ß Value Percentage of node Maximum Range
2 (LOS) 20 % 100 m
3 (LOS) 20 % 100 m
4 (N-LOS) 20 % 60 m
5 (N-LOS) 20 % 25 m
6 (N-LOS) 20 % 15 m
Table 3-8: Beta value for Nodes for power calculation from and to backbone router
25 3.3.11 Routing Algorithm
The routing algorithm in this simulation is Dijkstra’s algorithm. After implement Dijkstra’s algorithm in topology, several sensors connected to BR directly with bad link cost are discarded from BR according to the condition to the table: 3-4 in chapter 3. To illustrate the topology with Dijkstra’s algorithm in this simulation, a root-tree is generated according to this simulation. In appendix F an statistical results are given in tabular form with respect to each backbone router.
3.3.12 Cell Partition
In this section 240 sensor-nodes partitioned into eight cells and each cell has one backbone router. The topology of this wireless sensor network is based on the previous topology added each backbone router individually. In this scenario, the 240 nodes are partitioned into eight cells simultaneously, and each sensor-node should have connection to and only have connection to one cell. Meanwhile, the average number of sensor-nodes in each cell is 30, but this average number is an exception; if the number of sensor-nodes in one cell doesn’t exceed 48.
3.3.13 Partition Technique
The main focus of this technique is to keep balanced partition of every sensor-node into each cell, moreover make sure that the partitioned sensor sensor-node should not be belong to any other cell in the network.
The technique which is mentioned in this report is done on the basis of generations in a tree network with 8 back-bone routers.
3.3.14 Result of partition
After the topology with partition into eight cells, the number of sensor-nodes in each cell is:
Cell Number The Number of sensor-nodes 1(with BR1) 35 2(with BR2) 34 3(with BR3) 32 4(with BR4) 26 5(with BR5) 30 6(with BR6) 34 7(with BR7) 22 8(with BR8) 27
26
Below is the Statistical Result of Rout Cost (RC) in each Cell
Cell Number Max. RC Min. RC Average RC
1(with BR1) 2 1.00 1,267 2(with BR2) 2 1.00 1,196 3(with BR3) 2 1.00 1,312 4(with BR4) 2 1.00 1,212 5(with BR5) 2 1.00 1,341 6(with BR6) 2 1.00 1,329 7(with BR7) 2 1.00 1,054 8(with BR8) 2 1.00 1,237
Table 3-10 : Statistical Result of Router Cost in each Cell
The appendix G shows the complete tables which are showing the results of topology in each cell, the 240 sensor-nodes; Node 1-8 are Backbone router which is also the cell number. Field device numbers are starting from 9 and ended at 248.
27
4 Scenarios and Simulations
4.1 Introduction
This chapter will describe the simulation of different scenarios, which includes an overall simulation of 240 nodes, some special cases with different numbers of backbone routers and sensor nodes, 2 different industrial scenarios and some faulty scenario.
4.2 Simulation of 240 nodes
After the initial deployment of the node and mathematical calculations as per described in the chapter 3, the nodes are made to contact with each other without the obstacles in the network.
And during that simulation there are no backbone routers added, only nodes were configured as sources and destinations to find the path i.e. to make a neighbor table for each node with respect to other node in the network.
The whole process of making neighbor is divided into the following steps Step # 1 Distance between the nodes.
Step # 2 Calculate the Rx power at each node. Step # 3 Links Quality.
Step # 4 Links Cost.
4.2.1 Distance between the nodes
The distance between the nodes is calculated by equation 7 as per described in section 3.1.1, each node distance has been estimated with every other node, in order to get the orientation of all the nodes in the network.
4.2.2 Calculate the Rx power for each node
Rx Power has been calculated by equation 1 in section 2.2.1, in this step every node’s power has been calculated based on distance estimated by equation 7, the factors which have been used in this Rx power calculations are, Reference distance (do = 1m), attenuation factor, Gaussian random variable with standard deviation as 10 and mean as 0 and the received power at reference distance is -56 dBm. After finding the receive power an another estimation has been made to make a uniform distribution of power i.e. to add a random variable with a value of -4,4. As described in section 3.3.1.
28
4.3
Sub-Scenarios
In this section, some new simulation will be made, and this time 240 nodes scenario, is further broken down into 5 more scenarios listed below;
30 sensor nodes connected to each backbone router (BR1-BR8) respectively.
60 sensor nodes connected to each backbone router (BR1-BR8) respectively; and with a combination of 2 backbone routers.
120 sensor nodes connected to each backbone router (BR1-BR8) respectively; and with a combination of 4 backbone routers.
160 sensor nodes connected to each backbone router (BR1-BR8) respectively; and with a combination of 4 backbone routers.
200 sensor nodes connected to each backbone router (BR1-BR8) respectively; and with a combination of 8 backbone routers.
The motivation for this new simulation is to investigate average number of hops, maximum hops, minimum router cost, average router, and maximum router cost with the change in number of sensor nodes and backbone routers.
4.3.1 Sub-simulation 1:
30 sensor nodes connected to each backbone router (BR1-BR8) respectively.
In this simulation, the first 30 sensor nodes are selected out of 240 nodes, and later nodes are connected with 8 backbone routers (individually) at different time, located at different locations, as shown in figure 3-12.
A detail analysis has been made for each backbone router, and the factors which are calculated between router and sensor nodes are, average number of hops, maximum hops, minimum router cost, average router, and maximum router cost. The detail analysis with comparison will be shown in later chapter.
29
Figure 4-1: Top-to-bottom:Top: 30 nodes partition with backbone router 1 and 2 Bottom: 30 nodes partitioned with backbone router 3 and 4 individually
Figure 4-2: Top-to-bottom:Top: 30 nodes partition with backbone router 5 and 6 Bottom 30 nodes partitioned with backbone router 7 and 8 indivually.
30 4.3.2 Sub-simulation 2:
60 sensor nodes connected to each backbone router (BR1-BR8) respectively; and with a combination of 2 backbone routers.
This simulation comprises of two different steps,
Step 1, has been done in the same way, as described in section 4.3.1, the only change has been made is number of sensor nodes i.e. first 60 sensor nodes from 240 nodes.
Step 2, In this section, a difference has been made, by adding combination of 2 backbone routers at a same time over different locations, and the calculating factors are same as defined in section 4.3.1., below are the figures showing the architecture of network topology.
Figure 4-3 : Top-to-bottom: Top: 60 nodes partition with backbone router 1 and 2 Bottom 60 nodes partitioned with backbone router 3 and 4.
31
32 4.3.3 Sub-simulation 3:
120 sensor nodes connected to each backbone router (BR1-BR8) respectively; and with a combination of 4 backbone routers.
This simulation has been also broken down into 2 steps,
Step 1, is similar to the section 4.3.1, but with increased number of nodes.
Step 2, in this step, 8 backbone routers are broken down into 2 equal chunks, i.e., 4 backbone routers on the each side of the simulation area,
33 4.3.4 Sub-simulation 4:
200 sensor nodes connected to each backbone router (BR1-BR8) respectively; and with a combination of 8 backbone routers.
This simulation comprises of, 2 simulations, and divided into the steps.
Step 1, similar to 4.3.1 but with 200 sensor nodes, and in step 2, 200 sensor node simulations has been done using all 8 backbone routers atthe same time as shown in figure 4-7 below.
34
35
4.4 Obstacle based scenario
Since the simulation has to be done for the multi-path fading environment,
therefore an obstacle based scenario has been created for 240 nodes, and different number of simulations has been performed for the analysis purpose, the main theme for this simulation is to ensure the robustness of the sensor network topology for an obstacle based congested network.
Below is the figure describing the implementation of obstacle within the network, the dots below showing the sensor nodes while straight line showing the obstacle between the nodes.
The nodes are generated as per described in the previous chapter, while the obstacles have been created by using the very famous straight line formula, i.e. Y=mx+b, below is the maximized picture, which will show the nature of communication between the nodes.
Here in the figure, there are 2 sensor nodes, a solid line define the obstacle between the nodes, while the dotted line shows the communication breakdown between the nodes.
4.4.1 Obstacle based simulation
In this simulation ,obstacles are introduced , and the communicate is tested during the obstacle presence between the sensor nodes , it has been observed during the simulation that some nodes (links) get broken and found another path , and get connected with some other coordinator , or with the same coordinator but with different path.
36
Below is the tree showing the topology orientation during the simulation of the network.
Figure 4-9 : Partitioning of nodes between pan-coordinator after obstacles
The detail mathematical analysis with different factors including average number of hops, maximum hops, minimum router cost, average router, and maximum router cost, shall be discussed in the later chapter.
4.5 Redundant path simulation
There are three paths that a node can find to connect to backbone routers for sending data to Network Manager. Among these three paths two paths are within the cell (Basic and Intra Cell) and one is in between cells (Inter Cell).
The three paths are: 1) Basic Route 2) Intra Cell Routes 3) Inter Cell Routes.
Basic route are those which is used to connect to backbone router by any nodes in normal operating conditions. Whenever there is any obstacle in the Basic route the node will try to find other route inside its own cell first, i.e. Intra route, but if it is unable to find any routes in Intra route then it will try to connect to other Cells, directly with the backbone router or through nodes, which will have minimum routing cost.
1) Link failure
I. 10 percent of links in all cells fail. II. 20 percent of links in all cells fail. III. 30 percent of links in all cells fail.
37
2) Node failure
I. nodes failure in all cells II. 8 nodes failure in all cells III. 10 nodes failure in all cells
3) Backbone Router failure
I. Backbone router 2 failure
II. Backbone router 1 and Backbone router 2 failure
III. Backbone router 1, Backbone router 2 and Backbone router 5 failure The statistical analysis for the above defined scenarios will be defined in the next chapter.
4.6 Simulation for Industrial scenario
In this section, two industrial scenarios of WSN are illustrated in detail, and these two scenarios are: Scenario one, 30m X 30m, 30 nodes with 2 backbone routers in an area, which can be considered as quantities of nodes in a small room;
Scenario Two 40m X 250m, 100 nodes with 3 backbone routers in an area, which can be considered as quantities of nodes distributed along the center line in a narrow area.
4.6.1 Simulation methodology
The simulation of these two scenarios is based on the previous technology, mentioned in report version 1.2. But several mathematical models and coding- technology should to be modify according to the special requirements of these two scenarios. Referring to Shadowing Model, the random variable X is changed to X ~ N (0, 15), and the value of beta between every two nodes is set as β n-n
And the nodes in topology are FFDs, so each node needs to set either actuator, or sensor, or motor randomly. The percentages of actuator, sensor and motor are 18%, 52% and 30% respectively.
ß Value Percentage of node Maximum Range
2 (LOS) 10 % 100 m
3 (LOS) 10 % 100 m
4 (N-LOS) 20 % 60 m
5 (N-LOS) 30 % 25 m
6 (N-LOS) 30 % 15 m
38
Because the actuators are in the topology, the actuators should have the priority during the partition phase, and then keeping the number of nodes in each cell can be considered as the second priority.
Scenario One
Deployment of nodes and backbone routers
Figure 4-10 : Scenario 1: Random deployment of nodes
In this deployment 60 nodes are distributed in area 30m X 30m, and two
backbone routers are added into the topology: Backbone Router One (15, 0) and Backbone Router Two (15, 30).
The result of partition
In scenario one, the topology is partitioned to two cells, and one backbone router manages one cell. After partition, the number of nodes in cell 1 is 26, and the number of nodes in cell 2 is 34.
39
Scenario Two
Deployment of nodes and backbone routers
In this deployment 100 nodes are distributed in area 40m X 250m , and three backbone routers are added into the topology: Backbone Router One (62.5, 40) and Backbone Router Two (125, 40), and Backbone Router Three (187.5, 40).
Figure 4-12 : Deployment of Nodes
Figure 4-13 : Deployment of backbone router
The result of partition
In scenario two, the topology is partitioned to three cells, and one backbone router manages one cell. After partition, the numbers of nodes in cell 1, cell 2 and cell 3 are 20, 32, and 48 respectively.
40
41 0 20 40 60 80 100 BR 1 BR 2 BR 3 BR 4 BR 5 BR 6 BR 7 BR 8 Ce ll N etw o rk Per ce n tage o f N o d e s
240 Nodes
1 Hop Routes Multiple Hop Routes5 Findings and analysis
This chapter will describe all the analysis and findings which have been made during the simulation as described in previous chapters, the chapter will mainly discuss comparison between Single cell and multiple cells in a network, with different aspects, i.e. Reliable versus unreliable links, Max hop count versus Average hop count and etc.
5.1 Single Mesh and Multiple cell based network
As per it has been already discussed, that this project is for an implementation of an efficient network, which can work effectively well in multipath fading
environment. And provide least possible hops for data communication between different numbers of nodes.
The scenario here has 240 nodes altogether, and 8 backbone routers, different simulations has been performed for the measurement of different factors, The factors which has been measured are described in the coming sections.
5.1.1 Reliable and unreliable links
There are some certain criteria for finding the reliable and unreliable links, the nodes will assumed to be disconnected from the links if the link quality becomes worst as per discussed in section 3.3.1 equation 12.
The above table describes the link quality between the Nodes and backbone routers, as it has been described, that there are 8 back bone routers and 240 nodes, therefore for this analysis, there are 9 different simulations has been performed, 8 simulation with different backbone routers at different positions and 240 nodes, and the final one is with all backbone routers activated together for the
communication link between nodes and routers.
42 0 0,5 1 1,5 2 2,5 3 3,5 B R 1 BR 2 BR 3 BR 4 BR 5 BR 6 BR 7 BR 8 C e ll N etw ork H o p N u m b e r
240 Nodes
Maximum Number of Hops Average Hops 0 0,5 1 1,5 2 2,5 BR 1 BR 2 BR 3 BR 4 BR 5 BR 6 BR 7 BR 8 Ce ll N etw o rk R o u te Co st240 Nodes
Average…unreliable links, and in the simulation it has been noted that, until unless, there is a single backbone router operating, the link quality was worst, but as all the
backbone routers are activated, the link quality exponentially get improved, and the unreliable links also get reduced.
The main reason of getting exponentially improvement in the last simulation with 8 backbone routers is that, most of the nodes get connected by different routers with a single hop which nearly impossible to achieve by using single backbone router.
Figure 5-2: Route Cost Comparision
Figure 5-2 and 5-3, is giving proof, that until unless there is single backbone router operating at different positions individually, the hop counts will always be worst, but as soon as the number of backbone routers get increased in operation, the hop count greatly reduced and link quality will automatically get improved. In the end it is to be noted that this entire simulation has been done without any
obstacles in the network.
43 0 20 40 60 80 100 BR 1 BR 2 BR 3 BR 4 BR 5 B R 6 BR 7 BR 8 Ce ll… Per ce n tage o f N o d e s
Obstacle 240 nodes
1 Hop Routes Multiple Hop Routes 0 1 2 3 4 5 6 N u m b e r o f H o p sObstacle 240 nodes
Maximum Number of Hops 0 0,51 1,52 2,53 R o u te Co stObstacle 240 Nodes
Average RCFurther detailed results for this simulation can be found in Appendix F and Appendix G.
5.1.2 Obstacle based Reliable and unreliable links
This section is an extension of previously explained scenario in section 5.1.1,the only amendment which has been done in this section is introduction of obstacles in order to make wireless communication near to the real world , and it is also assumed that this will give a more realistic picture of hop count and links unreliability.
The above figure explains the link quality of the network, with respect to the obstacles in the network, as it has been obvious, that the communication becomes much worst when there are obstacles and single router operating in the network, as shown with the brown line in the graph.
But the link quality of the network, get improved, when there are more routers working together, i.e. all the 8 backbone routers operating at the same time. And it has been proved by the last simulation in which blue line exciding the brown, and shows the stability in the network.
Figure 5-4: 1 Hop Routes vs Multiple Hop Routes
Figure 5-5: Route Cost Comparision Figure 5-6: Maximum Number of Hops Vs Average Hop
44
operating at different positions individually with obstacles in the network, the hop counts will get even more worst as shown in figure 5-6, but as soon as the number of backbone routers get increased, the hop count greatly reduced and link quality will automatically get improved. This means that there will be more redundant paths operating in the network.
Further detailed results for this simulation can be found in Appendix H.
5.1.3 Sub-simulations for 60,120,160 and 200 nodes
In this section the sub-simulation of different sensor nodes will be analyzed along with the combination of different number of backbone routers as described in section 4-3; the main reason behind this simulation, to analyze all the possible conditions and combinations of sensor nodes and backbone routers, and conclude the results in term of Hop count and links reliability. Below are the graphical analysis of both hop count measurement and link reliability.
45 0 1 2 3 BR 1 BR 2 BR 3 BR 4 BR 5 BR 6 BR 7 BR 8 Ce ll 1& 2 C e ll 3& 4 Ce ll 5& 8 R o u te Co st
60 Nodes
Average RC 0 0,5 1 1,5 2 2,5 BR 1 BR 2 B R 3 BR 4 BR 5 BR 6 BR 7 BR 8 Ce ll… Ce ll… R o u te Co st120 Nodes
Average RC 0 0,5 1 1,52 2,5 BR 1 B R 2 BR 3 BR 4 BR 5 BR 6 BR 7 BR 8 Ce ll 1,2, 3& 4 Ce ll 5,6, 7& 8 R o u te Co st160 Nodes
Average RC 0 0,5 1 1,5 2 2,5 BR 1 BR 2 BR 3 BR 4 BR 5 BR 6 BR 7 BR 8 Ce ll… R o u te Co st200 Nodes
Average RC Figure 5-8: Sub-simulations for the hop analysis46 0 0,5 1 1,5 2 2,5 BR1 BR2 BR1 and BR2 Maximum Number of Hops Average Hops 0 20 40 60 80 BR1 BR2 BR1 and BR2 1 Hop Routes Multiple Hop Routes 1,45 1,5 1,55 1,6 1,65 1,7 1,75 1,8 BR1 BR2 BR1 and BR2
Average RC
Average RC5.2 Simulations for Industrial scenarios
In this section, 2 new simulations will be discussed, which are based on the Industrial environment, and has been already described in section 4.6;
The analysis which have been shown here are hop count analysis and the link reliability analysis.
5.2.1 Industrial Simulation 1
Figure 5-10 : Simulation for Industrial scenario 1
In this simulation the Backbone routers are fixed at a particular location, and altogether there were 60 nodes in the network with an area of 30X30m; and it has been observed with two backbone routers used at a time, the number of reliable links greatly increased with slight improvement in average number hops. Further detailed results for this simulation can be found in Appendix N.
