II. Mobile ad-hoc Networks – A brief introduction
2.1 Mobile ad-hoc Networks
2.1.1 MANET fundamentals…
Different protocols are at work for any communication network to be able to offer functionalities to users, and the MANET networks have a strict set of requirements and characteristics for protocol designs, each with the aim of achieving a specific step that allows the next ones to function efficiently to reach the final result.
To formulate the different terms commonly used in ad hoc networks and will be adopted throughout the research, definitions and concepts are introduced as follow:
- Node
Each of the devices participating in the network, agents with computational capabilities with diverse characteristics but a common wireless communication technology that allows it to be part of the mobile ad hoc network.
- Link
Assuming the existence of wireless communication technology in two or more nodes positioned at a distance from each other, a link is defined as the interconnection or communication between the nodes.
- Neighbor
Nodes located at a close range of communication from each other, a general term would be nodes located at a distance of jumps (n) from each other capable of direct communication among neighbors of degree n.
- Distance
Refers to the amount of links that need to be created for the information to be able to reach the destination, measured in number of jumps.
8
- Message
The information transmitted among nodes through the different links, the size of message and transmission speed depends on the characteristics of the wireless technology used by nodes.
- Routing
Refers to the paths through which nodes can communicate using a single or series of messages sent through different links or intermediate nodes. Due to the mobility of nodes in the network, the paths can change, appear or disappear.
The figure below shows the elements of an ad hoc network through different nodes and the communication links in between. Security, quality of service, power consumption management are all current challenges in the development of solutions for mobile ad hoc networks.
Figure 1: Mobile ad hoc Network representation 2.1.2 Access to communication
Wireless nature is an essential part of ad hoc networks and there are no dedicated transmission channels for the communication between nodes. Electromagnetic waves are propagated in the surroundings with the objective of creating the communication among nodes in the network, the absence of dedicated channels leave the waves open to elements capable of causing problems such as attenuation, interferences or even blocking of the signals.
These problems impose limitations that require especial control mechanisms to help reduce the impact of the same in the transmission process. The attenuation is the relation between the transmitted and received power, in a free space transmission scenario follows the inverse-square law which states that the received power of transmitted electromagnetic waves is proportional to the inverse of the square of the distance from the transmitted source.
𝑛𝑛𝑛𝑛𝑡𝑡𝐼𝐼𝑛𝑛𝐻𝐻𝐼𝐼𝑡𝑡𝐼𝐼 ∝ 𝑑𝑑𝑒𝑒𝑠𝑠𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑒𝑒1 2 (1)
Node Transmission area Link (wireless)
9
The presence of different obstacles and materials in the environment deteriorates the received power even more, requiring a higher transmission to overcome the loss.
At the same time, there is a very high number of different transmissions active in the environment sharing the same transmission medium, the different devices with wireless transmission capability are governed by different techniques to control the access to the shared channel and this way avoid the loss of information if they all were to access the channel at the same time.
In real world application, the devices that make part of the network are handle around every day common objects (e.g., furniture, walls, windows, etc.) that present obstacles for the arriving signal, not only signal block can occur, but other phenomena such as reflection, diffraction, dispersion are all a reality cause by the intermediate physical objects.
A number of techniques and methods had been developed to reduce the negative impacts of the different problems and make a best effort to solve them. Medium Access Control (i.e., MAC) is one of the techniques developed that aims to control the time when a device can transmit using the shared channel, and is defined by the current protocol in use Wi-Fi IEEE 802.11〔8〕
for device communication in ad hoc networks, same as the one used for the development of the current thesis.
2.1.3 Routing protocols in MANET networks Two groups can be defined as follow:
- Pro-actives
Different works were presented for this type of protocols and their functionality showed to be adequate for networks with a reduced number of nodes. The nodes in the network maintain a table with the information of the existing routes beforehand, a certain degree of previous knowledge is present since the information is created even before is needed, this characteristic makes it not suitable for large networks because the maintenance of the information and the constant sending of update messages create a high overhead in the system. Examples of this kind of protocols were presented in Destination Sequenced Distance-Vector 〔9〕and more recently in Optimized Link State Routing 〔10〕.
10
- Reactives
As the name implies, this type of protocols search for the communication routes at the time a node wants to establish communication with another node. Using the flooding approach, the search floods the network with the search query or message to establish a communication route, the route can be kept alive through different mechanisms for a certain period of time to overcome the changes in the topology of the ad hoc network as presented in the work 〔11〕.
Besides the two above mentioned groups, there are a variety of other protocols that make use of different characteristics and take advantage of the different positive features of the approaches, the hybrid solutions present a combination of pro-active and reactive characteristics, others focus on location-based approaches, using the geographical position through GPS to send the messages to the closest location to the destination that needs to be reached, energy aware approaches distribute the routing load among all nodes in an evenly manner and the multicast protocols that create and arrange groups to perform the routing tasks.
2.1.4 MANET applications
The characteristics of ad hoc networks make it suitable for situations where a rapid deployment of a network is needed and previous communication infrastructure does not exist. Emergency response systems, military operations, sensor data collection and monitoring represent areas of application for ad hoc networks. With the increasing popularity of sensors usage, governments, companies and privates are increasingly interest in benefiting from the advantages of data collection, measurements and monitoring of goods and spaces. Smart cities look to provide efficient services while saving valuable resources such as energy, reduction of CO2 emissions, waste management and others, the agricultural sector relies on technology to monitor, improve and maximize crop production, forests are equipped with a number of sensors to prevent wildfires or control the wildlife well-being as well as study behaviors and changes of the same.
In the case of emergency response, many natural disasters of high scale (e.g., earthquakes, tsunamis, flooding, etc.) leave the communication infrastructures completely useless, ad hoc networks capabilities allows rescue teams to deploy communication networks formed by the devices with computational capabilities itself, aiding in the search, repair and recovery of resources made unavailable following the collapse of the traditional communication network.
11
2.2 Simulation and parameters
Due to the dynamic nature of MANET networks, is difficult to design and develop tests in real world scenario, a number of devices with wireless capabilities populating the network should move along an environment with variable dimensions following patterns and different range of velocity. Hence is very common the use of simulators in researches related to mobile ad hoc networks due to the capability of providing behavior models for the elements that constitute the network, characteristics such as transmission, routing protocols, message traffic and mobility are all handle in a controlled test environment.
Throughout the years, scientists and academics have been discussing the reliability and credibility of the use of network simulators, 〔12〕and 〔13〕consider the tests made using network simulators provide significant validation of a particular proposal, as long as the parameters used during the tests are correct and adequate to the requirements of the system.
Although the most reliable results will only be obtained through real world experiments, the high inversion of capital to design the scenario and acquire the necessary devices for testing, besides the amount of time required to reach the optimum settings and without the guarantee the expected result will be reached, make the real world experimentation impractical and high risk.
The simulations performed in the research and presented in chapter 5 and 6 of the document, takes advantage of the capabilities offered by network simulators to provide a controllable environment. A range of 20 to 200 services is used in an area sized 800m X 800m, the nodes move at different speed and the response of the protocol is tested through different experiments with the adequate parameters settings. A real world experiment would require a large investment in capital and effort, factors that make it not possible to be developed for this research.
2.2.1 Network simulators
A variety of network simulators had been developed and used in many previous works related to the field. GloMoSim 〔14〕, OPNET 〔15〕〔16〕and ns-2 〔17〕share the common characteristic of being based on discrete event simulations. Meaning the behavior of the nodes in the network is modeled through the process of events occurring in the network, in the order they occur. The events are a response of the successes in the ad hoc network, such as sending message, receiving message, message collision; these events can also be generated by the user
12
through the scripts that define the behavior of the simulation. Besides, the simulators also provide models for the physical layer transmission, access and routing protocols, network traffic and mechanisms to generate and simulate the mobility of nodes inside the network.
- GloMoSim
Is a network simulator developed by the UCLA Parallel Computing Laboratory, it was developed as a wireless network simulator exclusively, the main advantage is its capability of executing in a parallel manner in different processors, reducing the execution time of a simulation. The simulator uses a parallel programming language called Parsec, and all the extensions and protocols to be included in the test environment should be written using this language. The company working on GloMoSim released the last version in the year 2000 before stop working on the open version of the software, later in 2011, a new commercially available simulator QualNet〔18〕was released based on the previous GloMoSim simulator characteristics.
- OPNET
A commercially available network simulator developed by OPNET Technologies with the main advantage of integrating the modeling, simulation and network analysis processes, the process is performed in a graphical approach, making it very intuitive with a low learning curve, very popular in corporation use and educational application.
- ns-2
This is probably the most well documented network simulator since is an open source software and developed in a collaborative manner, the license is distributed under GNU GPLv2 and is widely used for educational and research purposes. Currently includes contributions from different research centers and universities, making it an ideal candidate for the experimentation of protocols related to ad hoc networks. The protocols and extensions are written in C++, while the configuration of the environment and events is done using Tcl language. The main drawback with ns-2 simulator is the scalability, when a large network is simulated, the necessary computational resources increases greatly. A new version of the simulator, NS-3 〔19〕is being developed to facilitate the use and maintenance of the system as well as the future extensions.
13
Table 1: Network simulator characteristics
GloMoSim OPNET ns-2
Language C / Parsec C++ C++
Environment Parsec Graphical Interface Tcl
Java Client None None Agent J
License Educational purpose Limited edition GNU GPLv2
Advantages Parallel execution Simple to use Widely use and documented
Drawbacks No Java client, Parsec language
No Java client Scalability
The ns-2 network simulator was chosen to develop the simulations required by the thesis and validate the proposal of the same through the results and comparison. The open source nature of this simulator allows for a faster and easier understanding of the functionalities and protocol extensions through the extensive documentation available. Through the Agent J extension for the java client execution, the standard libraries and UDP sockets as the communication mechanism, validates the resource and service discovery algorithm presented in chapter 5 through real implementation without the necessity of real device deployment, the algorithm was developed and tested using Java programming language and C++.
2.2.2 Parameters
To configure the test environment of a network simulator a number of parameters need to be set. The variation of the same have a drastic impact in the results obtained during the execution of the simulation.
Parameters related to the simulation tool, the environment characteristics and solution related can be distinguished while designing the solution. A great number of researches do not specify the configuration of the parameters followed during the experimentation, which makes it difficult to re-create the behavior and results for comparison between solutions.
〔20〕identifies different settings that need to be correctly configured in the simulator to obtain reliable results, type of antenna, transmission distance, MAC protocol, query queue management and propagation model. NS-2 specification indicates more accurate results when 2-Ray Ground propagation model, consisting on the use of a direct path and a reflected from the ground, is chosen for long range transmissions, as well as an Omni-directional antenna type.
14
Other parameters related to the simulation environment include the amount of nodes, mobility and topology of the environment. The mobility of nodes is limited to the topology and size of the scenario, where transmission T in an rectangular test area of dimensions (a,b) define the maximum amount of nodes 𝑛𝑛𝑒𝑒𝑑𝑑𝑚𝑚 through the equation:
𝑛𝑛𝑒𝑒𝑑𝑑𝑚𝑚 =√𝑑𝑑2T+ 𝑏𝑏2 (2)
〔21〕presents a proposal for constructing scenarios that allow for a more rigorous evaluation of MANET networks solutions. To perform the simulations in this research, the two metrics proposed by the work were selected and are described below:
- Average network partition
Refers to the average percentage of nodes in a network that are disconnected in a given period. If the value is high, the functionality of a protocol for mobile ad hoc networks will not be reliable, low values for this parameter guarantees the protocol is creating communication routes among the majority of nodes in a network.
- Average shortest path
Refers to the average shortest distance allowed between two nodes in the network.
Incorrect settings of parameters may create a scenario where a great number of nodes are located at a one hop distance from each other, creating wrong measurements. The average number of jumps necessary for messages to reach a relevant node in the network should be well configured.
〔22〕identifies some mobility models that are significant to the evaluation of protocols. The selection of a mobility model depends on the concentration of the application for the developed protocol. For the experimentation in the thesis, the Random Waypoint model was chosen as the mobility model, based on the movement of nodes with aleatory speeds and directions it provides a general overview of the characteristics of the proposed solution without the need of moving into more specific details of the application.
Standards for the evaluation of the resource and service discovery protocols such as the one presented in this work are difficult to find, the test configurations and different proposals comparison are limited, a best effort assessment and analogy was applied during the test phase.
15
The table below shows the different parameters common for the experimentation presented in chapter 5, and that were selected and set following the typical configurations for mobile ad hoc networks.
Table 2: NS-2 parameters configuration for simulation
Simulation Parameter Value
Type of antenna Omni-directional
Propagation model 2-Ray Ground
MAC protocol IEEE 802.11
Query queue 5 max.
Transmission range 100m
Transmission speed [11 , 54] Mb/s
UDP packet (max.) 1500 bytes
2.3 Conclusion
The chapter introduced the fundamental characteristics of MANET networks to provide a better understanding of the limitations and requirements of this type of networks as well as the applicable solutions that fit the wireless transmission, access control and routing protocols of the same.
A comparison of the different network simulators available is presented, and the characteristics and advantages of the chosen ns-2 simulator along with a reasoning to be the simulation tool for the present work is also introduced. The possibility of performing simulations using the Java client and communication libraries fit the requirements to evaluate the proposed protocol.
The different parameters that act and affect the simulation and validation of the same where introduced and discussed in the final part, along with the set of parameters selected for the experimentation of the proposed solution in this thesis.
16
III. Current state of service discovery
The chapter has the objective of introducing the different solutions and methods for service discovery currently functioning in the MANET networks field. The study of the previous works and researches provides a classification of the advantages, limitations and characteristics of the solutions. The investigations showcasing the features that want to be highlighted for the present work and the ones best represented in the area were selected in the reference examination process.
3.1 Service discovery
Service discovery is the process that allows the participants of a network to realize the different services available in their surrounding as well as obtaining the necessary information for posterior access and use of the services. There are two essential processes that can be distinguished in service discovery:
- Service publish
The participants offer their services to other nodes using a publish method announcing the characteristics of the services available.
- Search request
Potential clients looking for services send a search request with the characteristics of the service that will satisfy the requirements of the search.
As mentioned in the previous chapter, there is a blurred line between service providers and clients in the MANET networks, since the participants in this type of networks usually take both roles simultaneously.
3.2 Classification of current solutions
Service discovery problem in ad hoc networks gathered great attention from researchers, many solutions try to adequate the traditional infrastructure solutions to the characteristics of mobile networks. 〔23〕proposes a series of aspects considered to present a classification of the solutions.
The following sections of the chapter present details of the aspects and classification of the solutions relevant to the present work:
17
3.2.1 Use of directories
The category includes two different proposals, solutions that make use of dedicated nodes that hold the directories containing the information of the services available in the network, and the ones that are based on the dissemination of the information among nodes without the use of dedicated nodes as directories holder.
The use of directories had been used since the first solutions for service discovery came out, where service providers would register their services through an administrator in one or many dedicated machines in the network, and the clients would search in the dedicated machines
The use of directories had been used since the first solutions for service discovery came out, where service providers would register their services through an administrator in one or many dedicated machines in the network, and the clients would search in the dedicated machines