The Different Forms Of Network Coding

Published Date: 02 Nov 2017

Contents

Introduction:

The main purpose of any communication system is to transfer the data between source and destination (for both single cast / multicast) without loss of originality. One can establish the communication path either through wired or wireless connections depending on distance. In wired communications, the data transmission will be done using wired-based technology which may include direct path between the source and destination. Wired communication is packet-based or connection-orientated. Connection based communication rely on a dedicated communication channel being set-up for exchange of data / signalling between end-points. Most wired based communications are now packet-based and data is broken into individually routable packets transmitted over an IP infrastructure. A Signalling and management packet also exists with throughout the wired network, and allows the propagation of routes to direct traffic. Numerous packet based technologies Ethernet, ATM, MPLS, PPP, have evolved and compete to become more widely deployed. Packetized communication can be considered to exist at 7 layers with layers 1,2 and 3 being relevant to communication on the wire. Switching occurs at layer 2 this is typified by immediate communication with the neighbouring devices and no consideration to global addressing. Layer 3 is the level at which decisions regarding traffic route taken, based on packet addresses is made. Layer 1 is the physical wire or media employed. Wired communications offer far higher speeds, with a possible 10 Gb now easily possible on one cable, wireless (n) has approximately 300 Mbps maximum. Wireless allows free movement of clients and deployment of networks in difficult environment such as the battlefield, where wires couldn't be laid.

With the advent of Internet protocols, wireless communication has exploited the field of communication for long distance communication. In many papers, authors presented about the importance of wireless communication in present world because of its advantages and the routing protocols established in communication. The network topologies used in design of communication system accompanied with different routing protocols in forwarding the information. Introduce of Network coding increases the throughput and high data rate for wireless communication. Wireless communications are RF (radio frequency) based and fundamentally a replacement of layers 1 and 2 when used in a packet based infrastructure. Although recently some Layer 3 routed networks were initiated. Through strict standardization wireless Ethernet allows data to traverse the air, using one of the most extensive layer 2 protocols available 802.11 which has had 4 main revisions a,b,g and n. Many other protocols exist such as WiMax, GPRS and satellite protocols. In case of wireless communication there won’t be any communication channel or path established between the entities.

Security is far higher in a wired network, historically wireless has been a serious security flaw and only recently is WPA/2 deployed more widely to combat this inadequacy.

Wireless is not essential for the internet infrastructure, it is not fundamental to the internet, it is the ease of accessibility it provides which is its fundamental benefit, with most wireless devices on the periphery of networks. When compared to wired communication, wireless communication has exposed its advantages for large distance communications include security and cost effective under some regulations.

Network Addressing and Protocols:

Application Layer

Presentation Layer

Session Layer

Transport Layer

Network Layer

Data Link Layer

Physical Layer

Fig. OSI model

Back ground Research:

Communication systems:

Network coding has been revolution in the field of Wireless communications. Many authors were presented this statement in their writings. There are certain parameters need to look when one have dealt with communications systems as security, reliability, efficiency and throughput.

Coding:

At the beginning stage of mobile communications, the analog signals were used in transmission of information between mobile users which results in bandwidth consumption and less distance transmission. Thus results in the less number of users with maximum loaded traffic. Introducing digital signals for mobile signal transmission helps the service provider to attain bandwidth and power effective. This helps in increasing users in number and also concerned to information security. Coding has become an essential part of any communication system and results in lowering the channel bandwidth and improves the efficiency.

Types of coding introduced earlier in communication systems:

Source coding:

In general source coding is used to compress the information at transmitter before transmission. This results in lowering the cannel bandwidth which results in high throughput and high data rate. After source coding, the information can be transmitted through a medium and when the transmitted information is received successfully, the receiver should decompress the received data in order to retrieve the original message signal. This can be also referred as end-to-end coding because the encoding will be performed at the transmitter and the only at the receiver information can be decoded.

Channel coding:

Channel is the medium through which information can be transmitted between source and destination. Coding can be done at the medium that helps to send the information more secured known to be channel coding. In channel coding the original information will be accompanied with extra bits called as redundancy bits and helps to differentiate the original information and erroneous information bits. Even channel coding is also known to be end-to-end as this is applied between transmitter and receiver nodes.

Network Coding:

The drawback of coding schemes introduced at earlier stage of communication was end –to-end connection which may include delay and also if the link or route between transmitter and receiver failed may result in system failure. Hence in order to overcome this drawback, Network coding was introduced by Ahlswede [1]. In general traditional computer networks has store and forward method where the routers receives the packets and forward the data without modification. But introduction of Network coding made this method simple but very effective way. Using network coding packets from same stream or different streams arrived at routers are allowed to coded or mixed before forwarding.

Network coding can be applied at application layer, network layer or physical layer depending on type of application.

Different forms of Network Coding:

Algebraic network coding

Physical layer Network coding

Multiple –inputs-multiple-outputs (MIMO)

Description: http://www.ee.kth.se/~jinfeng/CCNCW/fig1.png

Figure 1 Communication without Network coding

Description: http://www.ee.kth.se/~jinfeng/CCNCW/fig2.png

Figure 2 Communications with Network Coding

Source: http://www.ee.kth.se/~jinfeng/CCNCW/

Intra-session Network coding:

Inter-session Network coding

Analog Network Coding

Construction of Network codes for Multicasting:

The linear information flow algorithm

Random Construction

Linear Network coding:

Linear network coding is method used to improve a network’s efficiency, throughput and scalability. It supports the detection of attacks and eavesdropping within the network. This technique uses Shannon Information theory as the basic principle on which Physical layer works. Unlike depending on the packets received, the system using linear network coding will combine several packets and then transmit them across same channel which results in maximum possible information flow in the network. Yeung et.al proved that linear coding is enough to achieve the upper bound in multicast problems in the paper entitled"Linear Network Coding" \cite{ 1176612}

Random network coding:

Random Network is defined as simple but powerful method in case of network coding\cite {1228459}. Using decentralized algorithm, optimal throughput is achieved in broadcast transmission. Linear combination of packets will be received by nodes in the network with the coefficient chosen from a Galois field. If the field is too large allows the linearly independent combination which results to obtain innovative information. Even though Random network coding is known for high throughput if any node obtain insufficient packets which may results in loss of original packets.

Literature Review:

In 2000, Ahlswede et al. initiated Network coding and published a paper entitled "Network Information Flow". \cite{850663} In this paper they have proposed coding techniques in the network employed at nodes which increases network capacity compared to normal routing and also able to prove the max-flow –min-cut theorem in case of Information flow which can be regarded as generalization of classical max-flow –min-cut theorem for commodity flow. In 1956, Max-flow-min-cut theorem for Commodity flow was proved independently by Elias et.al and Ford and Fulkerson in the paper called "An algebraic approach to network coding".\cite{ 935967 } Ahlswede et al. were able to prove the max-cut-min-flow for information flow that supports the generalization of this theorem for any commodity flow. Koetter and Medard extended the concept of Ahlswede by providing codes with simple and linear structures were enough to achieve Network coding for network. Later on, Ho et al. proposed concepts of achieving Network coding using random construction of linear codes and proved those A random Linear codes were sufficient in executing Network Coding in the paper "A random linear network coding approach to multicast". \cite {1705002} The usage of linear random codes were first realized by Lun et.al and presented in the paper entitled "On Coding for Reliable communication over packet Networks." \cite {1523665} Raymond W. Yeung in his paper "Network Coding: A historical Perspective" feels that works on Network coding has been increasing especially in practical scenario’s and as per the Yeung more than 80 papers were written on Network coding in each month. This statistics shows the advantage of Network coding in the field of Communications.\cite{5699900}

Network Topology:

Philip J Irving in his book "Computer Networks" defined Network Topology as the system through which different elements (such as links or nodes etc.) are connected with in a Network and also includes the mode of communication exists among them.

In general Network Topology can be shown in two forms:

Physical Topology

Logical Topology

Physical Topology:

Physical topology is defined as the interface of different components among the nodes in Network. This component includes cable installation and device location.

Logical Topology:

Logical topology shows the process includes in data flow within Network irrespective of the type of their design.

Topologies for different Networks may be identical irrespective of some factors such as transmission rates, physical interconnections, distance between the nodes and signal types. (Irving, 2003)

Types of Network Topology: [3]

Mesh Topology

Star Topology

Bus Topology

Ring Topology

Tree Topology

Mesh Topology:

In Mesh Network, there are many connections existing between the nodes and in order to communicate between nodes a path has to be established first. These networks are complex in design and made to be flexible when there is a link failure in the network.

Description: http://t1.gstatic.com/images?q=tbn:ANd9GcRFpGNUR-KOJIRJSlD0I8153zg_tG72cPo_xHBhy-rQ5gloxWQZ

http://ecestudyaid.blogspot.co.uk/2012/07/different-kids-of-network-topology-in.html

The Mesh Topology can be further classified into two types:

Full Mesh Network

Partial Mesh Network

Full mesh topology:

In a Mesh Network, if each node is directly connected to other nodes in the network. Then such types of mesh network are called as full mesh Network. Despite of high redundancy full mesh networks are not preferred because of their high cost. In this case if there is any case of node failure then network traffic can be forwarded through other route. Full mesh networks can be used as backbone networks. (Irving, 2003)

Partial mesh topology:

The networks in which some of the nodes are connected to all other nodes in the network where some of nodes are connected only few nodes (one or two nodes) in the network are called as partial mesh networks. These networks are less expensive and produce less redundancy than full mesh topology. Partial mesh topology is usually used in peripheral networks associated to a full meshed backbone. (Irving, 2003)

Advantages and disadvantages of Mesh Networks: [4]

Advantages:

Fault Tolerance

Easy Troubleshoot

Guaranteed Communication channel capacity

Disadvantages:

Difficulty of installation and reconfiguration

Cost of maintaining redundant link

Star Topology:

In star topology all the nodes in the network are connected to a central node called as Hub. Each node in the star network communicates with central node that resends the information to all other nodes or only to the destination node. The central node in the star topology can be active or passive depending on the communication.

Description: http://1.bp.blogspot.com/-f1_JocDm57Y/UA-rnslcCCI/AAAAAAAAAIA/jVQ9wo3hIgE/s1600/star+topology.jpg

http://ecestudyaid.blogspot.co.uk/2012/07/different-kids-of-network-topology-in.html

Advantages and disadvantages of Star Networks are: [] [4]

Advantages:

Robustness

Performance

To diagnose the network faults is easy

Single computer failure do not necessarily bring down the whole net

Several cable types can be used with the hub

Disadvantages:

Central hub fails, the whole network fails to operate

Rebroadcast or switch network traffic.

More Cost.

Bus Topology:

In bus topology, every node in the network is connected to a single cable. The communication among the nodes will be occurred using the cable that connects all the nodes. The cable should be terminated at both ends else there will be interference of signals along the cable.

Description: Bus Topology

http://www.webopedia.com/DidYouKnow/Hardware_Software/Networking/bus_topology.html

Advantages and disadvantages of Bus Topology:

Advantages:

Simple to construct

Less amount of cable

Easy to extend

Disadvantages:

Single break in wire will make complete system shut down

Difficult to identify the Break down

Ring Topology:

In Ring topology all the nodes in the network are connected in a closed loop. In order to communicate with the destination node, source node transmits the information to the next adjacent node and the process goes on until it reaches the destination.

Description: Ring Topology

http://www.webopedia.com/DidYouKnow/Hardware_Software/Networking/ring_topology.html

Advantages and Disadvantages of Ring Topology:

Advantages:

Robustness

Higher capacity

A fair method of access

Disadvantages:

High Cost

Tree Topology:

Tree topology is also known as Hybrid Topology because it involves both star ad bus topologies. In a tree network, a group of nodes in star networks are connected to a linear bus cable.

Description: Tree Topology

http://www.webopedia.com/DidYouKnow/Hardware_Software/Networking/tree_topology.html

Advantages and Disadvantages of Tree Topology:

Advantages:

Division of complex networks

Easy to extend

Disadvantages:

Central hub failures may lead to network failure.

Description: network topology image

http://www.atis.org/glossary/definition.aspx?id=3516

Various attacks on wireless sensor networks are:

Eavesdropping

Data modification / injection

Node comprising

Logical and Physical Addressing:

Physical addressing which is also known as Layer two addressing

The OSI seven Layer Model:

In early stages of computer networking the development was slow and in a patented fashion. But after the formation of networking bodies together under the name of International Standards Organisations (ISO) and introduced open system interconnections (OSI) model. This model basically classifies the communication process into several steps which are small and securely defined in form of layers. The OSI seven layer model is also known as ISO seven layer model.

Description: http://www.infocellar.com/networks/images/OSI-1.png

http://www.infocellar.com/networks/osi-model.htm

The seven protocols can be classified into three groups as High Level Protocols, Medium Level Protocols and Low Level Protocols.

High Level Protocol:

This group includes Session, Presentation and Application layer. A main objective of this group of protocols is to analyse the data by presenting, displaying and summarizing for the user or else to maintain the data so that user can easily understand.

Application Layer (Layer 7):

Application Layer supports the application and end user interface. At this layer Communication and Quality of service are identified. Both authentication and privacy assigned to user are considered at this layer. The constraints associate with data syntax is observed and mainly deals with application specific interface. Application services such as file transfers, email and other software services are the functions operated at Application Layer.

Presentation Layer (Layer 6):

The data received from hosts in different formats will be encrypted which will be differ from both existing formats results in network format which is an independent data representation and vice versa. The presentation layer is responsible for the resulted data format to be acceptable by Application Layer and also for providing free from compatibility problems which occurs due to layer formats and encrypting data. This layer is also referred as Syntax Layer.

Session Layer (Layer 5):

The main function of Session Layer is to establish, manage and terminate the sessions between applications of two communicating hosts. Set up connections, allocating coordinates of hosts to each other, information exchanges are the functions dealt at session layer.

Medium-level Protocols (Layers 3 and 4 - Network and Transport):

The transmitted data will be assembled into frames and packets. Flow control and error checking will be implemented at this level.

Transport Layer (Layer 4):

The transport layer deals with reliable service, error recovery, transport error detection, and information flow. Mainly, the information from the sender breaks into units called segments. This layer supports to layer 5 with transport service and supports reliability and flow control for transmission of information.

Network Layer (Layer 3):

The network layer consists of logical addressing and provides path selection and connectivity between the communicating hosts. Internet operates at this layer and deals with technologies include switching and routing and the hosts transfer the information through appropriate paths known as virtual circuits, which is used to establish path between nodes. In this layer, the data which has broken into segments are still divided into packets and thus used in transmission across the Internet. Headers and tailers are added with the information packets which is suitable for internet transmission. The important element of header is the presence of source and destination address and elements in tail is CRC which made the transmission more appropriate. Addressing, error handling, congestion control and packet sequencing are the functions associated at Layer 3.

Low-level Protocols (Layers 1 and 2 - Physical and Data Link):

This level of protocols mainly deals with conversion of data into electrical or mechanical format to surpass over the physical medium.

Data Link Layer (Layer 2):

The main function of data link layer is to transmit the information across a physical link that uses physical address. The data received from the network layer will be in the form of packets which are further divided into frames at Data Link Layer and transmitted over the implemented Network technology.

Data link layer is further classified into two layers:

Media Access Control (MAC) layer

Logical Link Layer (LLC) layer.

The MAC layer involves with the host on the network to access the data and also authorization of transmission to other layer whereas LLC layer is used to control frame synchronization, flow control and error checking.

Physical Layer (Layer 1):

This layer mainly deals with mechanical, electrical, procedural and functional specifications for activating, maintaining and deactivating the physical link between end systems. The data is further transmitted into bits and it also provides hardware means of receiving and sending the data on carrier.

The TCP/IP model:

TCP/IP model is defined as open standard model for internet, which is more appropriate than OSI model. It helps in data communications anywhere located at global with the speed of electricity.. The defence system of US designed a network that specifically used to communicate across the world at the time of world war and also nuclear war and the requirements to make internet as a wireless mesh networks aroused the idea of four layered model which is presently known as TCP/IP model and the layered model behaves same as in OSI model.

The TCP/IP model consists of following layers:

Application Layer

Transport Layer

Internet Layer

Network Access Layer

Application Layer

Transport Layer

Internet Layer

Network access Layer

Application Layer:

The designers of TCP/IP model created application layer to cover all the functions which are carried out by higher level layers of OSI model or even application- oriented protocols. Because of this reason Application layer is also referred as Process layer. The functions handled at this layer are issues of representation, high-level data protocols, dialogue control and encoding. This layer allows the data to be transmitted in proper format to next layer.

Transport Layer:

Transport layer is also called as Host-to-Host layer because of its property of connection establishment between two communicating hosts. This layer remains at same place in TCP/IP model when compared with OSI model. Transport layer converts the data into set of segments which are carried to next layer. There are some protocols used at this layer which helps the data transmission between the source and destination across the internet. Transmission control Protocol (TCP) and User datagram Protocol (UDP) are the protocols used at this layer whereas Transmission control protocol (TCP) used in connection oriented services that includes Domain Name services (DNS), Simple Mail Transfer Protocol (SMTP), File Transfer Protocol (FTP) and Hyper Text Transfer Protocol (HTTP). TCP is a reliable transmission protocol as it is a connection oriented service. On the other hand User Datagram Protocol (UDP) is used for connectionless services. This protocol allows the data to transfer in faster way but the only disadvantage with this protocol is of its unreliable nature. Mainly UDP is used for trivial FTP (Known as TFTP) and rarely in Domain Network Services.

Internet Layer:

The segmented data from the transport layer will be further converted into packets and transmitted over the internet to the destination independent of paths and the networks. The selection of the best possible route will be important criteria in this layer which is known as Path determination. Switching routers helps the data to place on link. Internet layer exhibits the same behaviour of network layer in the OSI model.

Network Access Layer:

Network Access Layer exhibits the property of both physical and data link layer of OSI model. Data encapsulation into frames and bits are the functions carried out at this layer. This layer is also referred as host-to-network layer because IP packets are need to transfer between the physical link between the devices.

Comparison of OSI model and TCP/IP model:

Similarities between the models:

1. TCP/IP model and OSI model are layered protocols which exhibits the advantage of layering.

2. Application Layer is common at both the models but application layer in the TCP/IP model exhibits the properties of Session and Presentation Layer.

3. Both models exhibits packet switched technology.

4. Transport Layer is similar in each model.

Differences between the models:

1. TCP/IP model is less complex and as it has few layers.

2. TCP/IP combines both the presentation and session layers into the application layer whereas these layers are different in OSI model.

3. TCP/IP combines both data link layer and physical layer into single layer called as network access layer.

4. OSI model is the basic model used to construct the networks.

Description: http://www.infocellar.com/networks/ip/Images/osi-dod-tcpip.gif

http://www.infocellar.com/networks/osi-model.htm

Network coding can be applied in various layers such as physical layer, Data link layer and Presentation layer In OSI model whereas it can be applied at Network access layer and Application layer when TCP/IP model is used.

Wireless networks:

Akyildiz and Xudong Wang in their book "Wireless Mesh Networks" published that the services developed with the evolution of wireless networks made the impact on the way of communication especially in Internet for the next generation./cite{1509968} The authors felt introduction of wireless mesh networks (WMNs) allows the users to use internet everywhere or anytime with low cost. In comparison with conventional wireless networks, WMNs are associated with minimum upfront investment performing self-organization and self-configured and also used to reduce the system complexity.

In WMNs, mesh routers and mesh clients are the intermediate nodes used in communications. Unlike in conventional wireless networks, the nodes in WMNs act not only as host but also as a router which helps in forwarding packets instead of other nodes with in its transmission range. The nodes in WMNs used to establish and maintain the connectivity in the network.

The advantages of WMNs over wireless networks are:

Low up-front costs

Simple Network Infrastructure

Robustness

Reliable service coverage.

In order to connect to wireless Mesh routers, nodes such as laptops, PDAs, desktops of conventional wireless network uses Wireless network interface cards (NICs) or even with directly connected to the wireless mesh router in WMNs system (for example: in case of Ethernet).

The drawback of WMNs system is the availability of medium access control (MAC) and routing protocol which results in non-scalability of network. For example: with the introduction of new hops or even with the addition of more nodes to the existing system throughput of system drops significantly. On the other hand, the available security systems for conventional wireless networks will be effective for some of the attacks in WMNs but they lack of broad mechanism to prevent attacks from different protocol layers.

In order to use the resources of conventional wireless networks in WMNs, both MAC and Physical layer has to be reconsidered or even in some cases new protocols has to be made.

Characteristics of WMNs: