Abstract: Multicasting is the ability of a communication network to accept a single message from an application and to deliver copies of the message to multiple recipients at different locations. Our research in Multicasting, as to design a Multicast Model, which provides reliability & scalability with best path for data delivery. Reliability means guaranteed Delivery of packets. Scalability means capability to serve growing needs .In this context, a few concepts of Proactive-Routing technique are used to make available this model in Infrastructure wireless also. Minimum Spanning path and combo-casting is used to reduce the cost & delay and thus to deliver the packets. The main characteristic of RSM2 model is, to provide complete multicasting, i.e. at the same time more than one node can act as sender. This model provides one-to-many communications as well as many-to-many communications. The goal of thesis is to design an algorithm describing the function and behavior of Multicast Model (RSM2).
TABLE OF CONTENTS
DECLARATION
CERTIFICATE
ACKNOWLEDGEMENTS
ABSTRACT
LIST OF FIGURES
LIST OF TABLES
LIST OF ABBREVIATIONS
CHAPTER - 1 INTRODUCTION
1.1 Background
1.2 Motivation
1.3 Problem Statement
1.4 Dissertation Outline
CHAPTER - 2 LITERATURE REVIEW
2.1 Multicasting
2.2 What is Multicast?
2.3 How to Define Cost of the link
2.4 Minimum Spanning Tree
2.5 Ack-Implosion Problem
2.6 IGMP
2.7 Flooding Vs. Broadcasting
2.8 Various Approaches to Reliable Multicasting
2.8.1 SRM
2.8.2 RMTP
2.8.3 Light-weight Reliable Multicast Protocol
CHAPTER - 3 SYSTEM DESIGN
3.1 Design Objectives
3.2 Different Multicasting Scenarios
3.3 RSM2 Architecture and Assumptions
3.3.1 Flat Design of RSM
3.3.2 Structure of an Echo Packet
3.3.3 Structure of the Data Packet
3.3.4 BUFFER MANAGEMENT
3.3.5 Management of NACK Buffer
3.3.6 DYNAMICS MANAGER
3.4 General Description
3.4.1 DESCRIPTION OF THE MODEL (RSM2)
3.4.2 Working of the Model
CHAPTER - 4 PROPOSED ALGORITHM : MCPA
4.1 Basis for the Proposed Algorithm - MCPA (Minimum Cost - Path Algorithm)
4.1.1 Proposed Algorithm for RSM2 -MCPA
4.1.2 Algorithm for RMTP protocol
4.2 Algorithmic Computations to the Network Topology
4.2.1 Exposure of RSM2 to the networks
CHAPTER - 5 IMPLEMENTATION OF MCPA ALGORITHM
5.1 SNAPSHOTS
CHAPTER - 6 COMPARISON OF RSM2 WITH RMTP
6.1 Graph shows the complexity of RSM2 vs. RMTP
CHAPTER - 7 CONCLUSION AND FUTURE SCOPE
REFERENCES
PUBLICATION
DECLARATION
I hereby declare that this submission is my own work and that, to the best of my knowledge and belief, it contains no material previously published or written by another person nor material which to a substantial extent has been accepted for the award of any other degree or diploma of the university or other institute of higher learning, except where due acknowledgment has been made in the text.
Signature
Ruchi Gupta
CERTIFICATE
Certified that Ms. Ruchi Gupta has carried out the research work presented in
this thesis entitled “A RELIABLE & SCALABLE MULTICAST MODEL (RSM2)” for the
award of Master of Technology from Mahamaya Technical University, Noida under my
supervision. The thesis embodies results of original work, and studies are carried out by the
student herself and the contents of the thesis do not form the basis for the award of any other
degree to the candidate or to anybody else from this or any other University/Institution.
illustration not visible in this excerpt
ACKNOWLEDGEMENTS
I believe that I would not be able to name everyone separately that they did for me, however, I would like to take the opportunity and express a few words of thanks to my guide, colleagues, friends and family.
Special thanks and gratitude to:
Mr. Pramod Kumar Sethy, Assistant Professor, K.E.C, for accepting me & carrying out research work under him. Your prolonged interest in my work & excellent guidance has shown me a way to pursue excellence & reach my goals. Your cool mind, patience & attitude towards excellence have been a constant source of inspiration for me.
Mr. Rahul Prakash, Assistant Professor, CS-IT Dept., Mewar University, Rajasthan, for your constant support, understanding, help, patience, care, affection and helpful discussions. I am indebted to you from the bottom of my heart.
Sanjiv Kumar, an Engineering Scholar, Mewar University, Rajasthan. I would like to express my best and special thanks for all of your support.
Words are insufficient to express my profound sense of gratitude to my parents & friends, whose encouragement & blessings gave me great physical & moral strength. I would like to thank one of my dearest friends, Shailendra Jaiswal, for the valuable support, care and encouragement.
I would like to express my gratitude to the Department of Computer Science & Engineering,
Krishna Engineering College, Ghaziabad , that gave me the possibility to complete this thesis.
ABSTRACT
Multicasting is the ability of a communication network to accept a single message from an
application and to deliver copies of the message to multiple recipients at different locations. With
the emergence of mobile users, many existing Internet-Protocols, including those with multicast
support, need to be adapted in order to offer support to this increasingly growing class of users.
Our research in Multicasting, as to design a Multicast Model, which provides reliability & scalability with best path for data delivery. Reliability means guaranteed Delivery of packets. Scalability means capability to serve growing needs .In this context, a few concepts of Proactive-Routing technique are used to make available this model in Infrastructure wireless also. Minimum Spanning path is used to reduce the cost & delay and thus to deliver the packets.
The main characteristic of RSM2 model is, to provide complete multicasting, i.e. at the same time more than one node can act as sender. This model provides one-to-many communications as well as many-to-many communications .The goal of thesis is to design an algorithm describing the function and behavior of Multicast Model (RSM2).
In RSM2 model, Dynamics Manager plays an important role. Dynamics Manager is the
specialised Machines, with network computational capabilities. Dynamics Manager is the main
focus of this model. Dynamics Manager’s functionality makes available this model to work in
both wired and wireless networks. Dynamics Managers act as listeners and calculator to perform
network computations.
LIST OF FIGURES
FIGURE 2.1 OVERVIEW OF MULTICASTING
FIGURE 2.2 ACK-IMPLOSION PROBLEM
FIGURE 2.3 FLOODING ON THE NETWORK
FIGURE 2.4 BROADCASTING ON NETWORK
FIGURE 3.1 STRUCTURE OF AN ECHO-PACKET
FIGURE 3.2 STRUCTURE OF DATA- PACKET
FIGURE 3.3 BUFFER LOOKS LIKE AS A STACK
FIGURE 3.4 PROCEDURE FOR OPTIMIZED FLOODING ALGORITHM
FIGURE 3.5 ECHO PACKET FLOWS THROUGHOUT THE NETWORK USING OPTIMIZED FLOODING ALGORITHM
FIGURE 4.1 NETWORK TOPOLOGY
FIGURE 4.2 COST-MATRIX CREATED AT THE SENDER’S SITE
FIGURE 4.3 SORTED WORK-MATRIX
FIGURE 4.4 PRIORITY-MATRIX
FIGURE 4.5 NETWORK TOPOLOGY AFTER NODE-D LEAVES GROUP-G3 AND JOINS GROUP G
FIGURE 4.6 PARTIAL COST-MATRIX CREATED AT THE DYNAMIC MANAGER’S SITE
FIGURE 4.7 NODE -‘A’ WANTS TO COMMUNICATE WITH ‘GROUP-G2’ AND ‘GROUP- G3’
FIGURE 5.5 DECIDING THE TOPOLOGY OF NETWORK AND ASSIGNING THE COST BY USER
FIGURE 5.6 DECIDING THE TOPOLOGY OF NETWORK AND ASSIGNING THE COST BY USER
FIGURE 5.7 FORMATION OF COST-MATRIX , SORTED WORK MATRIX & PRIORITY MATRIX
FIGURE 5.8 EACH NODE IS ASSIGNED TO THE SPECIFIC GROUP , AS ENTERED BY USER
FIGURE 5.9 SENDER NODE AND RECEIVER GROUP IS ENTERED BY USER
FIGURE 5.10 SENDERS(S) USE PRIORITY MATRIX TO SEND THE DATA TO DESIRABLE RECEIVERS(R)
FIGURE 6.1 GRAPH SHOWING RSM2 VS. RMTP
LIST OF TABLES
TABLE 4.1 COMPLEXITY ANALYSIS FOR MCPA ALGORITHM
TABLE 4.2 COMPLEXITY ANALYSIS FOR RMTP ALGORITHM
TABLE 4.3 NODES WITH THEIR GROUP-IDENTITY
TABLE 4.4 DYNAMICS MANAGER STORES THE INFORMATION OF NODES & COST OF EDGES AS MATRIX
LIST OF ABBREVIATIONS
illustration not visible in this excerpt
CHAPTER - 1 INTRODUCTION
1.1 Background
In the last few years, the Internet has changed from a pure scientific network to the basis of data communication in every-day life. The number of users grows still exponentially and has already reached the order of magnitude of tens of millions. The added spectrum and number of users introduce also new forms of communication into the Internet Communication, not just between two peers, but true group communication. The foundation for the group comm. in the Internet is the IP-multicast service.
Although most of the network connections that were used before multicasting have been known, were Unicast (one-to-one) connections. They were basic and can be used for reliable data transmissions back and forth between the two connected nodes. However, these connections are not appropriate for communications from one sender to multiple receivers (one-to-many) or for many senders to many receivers (many-to-many).
Multicast connections, particularly IP Multicast connections, may be more appropriate. IP Multicast connections may be particularly useful for one-to-many communications, as the sender need only transmit data packets once, and multicast enabled routers will copy the packets and send them to joined receivers. As the standard is implemented, there is no need for a sender to generate as many copies of the packets, as the number of receivers.
The need only is to know the Multicast-IP address. Multicasting is the ability of a communication network to accept a single message from an application and to deliver copies of the message to multiple recipients at different locations. Multicasting represents an efficient mechanism that implements point-to-multipoint communications1.
The two important issues in communication network are: Reliability (guaranteed Delivery) and
Scalability (serving growing needs). Our aim to design this model is to deal with these issues.
We use Proactive routing in this model, to route the packets. The three basic mechanisms of this Model (RSM2) are neighbour discovery, to discover and maintain connectivity with peers, and constructing minimum spanning path using adjacency matrix.
1.2 Motivation
Reliable multicast is required by many applications such as Multicast File-Transfer, shared white-board, distributed interactive simulation and distributed computing. These applications can potentially have several thousands of participants scattered over a wide area network.
Even though current multicast networks, such as IP multicast networks, provide efficient routing and delivery of packets to groups of receivers based on multicast group addresses, but they are lossy and do not provide the reliability needed by the above applications. Designing scalable approaches and architectures for reliable multicast for an efficient use of both the network and end-host resources, is a challenging task.
Local recovery approaches for reliable multicast, in which a network entity other than the sender aids in error recovery, have potential to provide significant performance gains in terms of reduced bandwidth and delay and higher system throughput. We are using here Receiver-Initiated NACK based Reliable technique and Active-Server based local recovery, to providereliability & scalability. Another thing that motivates me to design this model is, the nature of nodes to dynamically change their groups. For this purpose, IGMP protocol -II is applied in RSM2. Proactive Routing and combo-casting is used, to make available this model in Dynamic Environment.
1.3 Problem Statement
“TO DESIGN AN ALGORITHM DESCRIBING RSM2 MODEL”
There have been many models and protocols (SRM, RMTP...) have been introduced, to provide reliability and scalability in multicasting communications. The goal of the research is to design an algorithm describing Multicast Model (RSM2), which achieves scalability and reliability. Along with it, our model opt flat approach. The problem with hierarchical approach is that, every time when a receiver becomes the sender, entire hierarchy was changed. The hierarchical Model, RMTP, does not fit in a situation of, where, many nodes can send data simultaneously at same time. Another problem was to choose an effective Designated Receiver, to deliver the packets to all the nodes, under it. Our Multicast Model removes all the above stated problems and provides a cost-effective, delay-effective path to deliver the packets.
1.4 Dissertation Outline
The thesis is organized into seven chapters. The first chapter provides an introduction to the topic being discussed .This chapter gives a brief focus on the background material, studied to complete this work. This thesis is aimed at designing an algorithm, which describes our Multicast Model(RSM2).
The second chapter describes the literature review and related work in the field of Multicasting.
The third chapter describes the work done in this thesis. The work done is divided into 4 sections listing the purpose, general description and working of the Model.
The fourth chapter introduces a proposed algorithm known as MCPA (Minimum Cost-Path Algorithm).
The fifth chapter provides the snapshots after implementation of MCPA algotithm for describing RSM2.
The sixth chapter provides a comparative study of RSM2 with RMTP.
The seventh chapter concludes the thesis and provides recommendations for future work.
CHAPTER - 2 LITERATURE REVIEW
2.1 Multicasting
MULTICASTING provides an effective and efficient way of disseminating data from a sender to a group of receivers. Instead of sending a separate copy of the data to each individual receiver, the sender just sends a single copy to all the receivers. A multicast tree is set up in the network with sender at the root node and the receivers at the leaf nodes. Data generated by the sender ows through the multicast tree, traversing each tree edge exactly once 4. However, distribution of data using the multicast tree in an unreliable network does not guarantee reliable delivery, which is the prime requirement for several important applications, such as distribution of software, nancial information, electronic newspapers, billing records, and medical images.
2.2 What is Multicast?
Normal IP packets are sent from a single source to a single recipient. Along the way these packets are forwarded by a number of routers between the source and recipient, according to forwarding table information that has been built up by configuration and routing protocol activity. This form of IP packet delivery is known as unicast.
However, some scenarios (for example, audio/video streaming broadcasts) need individual IP packets to be delivered to multiple destinations. Sending multiple unicast packets to achieve this is unacceptable because it would require the source to hold a complete list of recipients. Multiple identical copies of the same data would flow over the same links, increasing bandwidth requirements and costs. Instead, data to multiple destinations can be delivered using multicast.
Multicast allows the source to send a single copy of data, using a single address for the entire group of recipients. Routers between the source and recipients use the group address to route the data. The routers forward duplicate data packets wherever the path to recipients diverges.
A multicast group identifies a set of recipients that are interested in a particular data stream, and is represented by an IP address from a well-defined range. Data addressed to this IP address is forwarded to all members of the multicast group.
A source host sends data to a multicast group by simply setting the destination IP address of the datagram to be the multicast group address. Sources do not need to register in any way before they can begin sending data to a group, and do not need to be members of the group themselves.
In the following diagram, S2 sends a single copy of its multicast data addressed to the multicast group. The group consists of hosts G1, G2 and G3. The data is duplicated at routers R1 and R3 to ensure that it reaches all the hosts that are interested in this multicast data. G4 and G5 do not belong to the multicast group, and hence do not receive copies of the data.
In the diagram as shown below,
- S indicates a host sending multicast data
- G indicates a host which may or may not be a member of the multicast group
- R indicates a multicast-capable router.
illustration not visible in this excerpt
Figure 2.1 Overview of Multicasting
Information about which parts of the network contain members of a particular multicast group is distributed as follows.
Hosts who wish to receive data from the multicast group join the group by sending a message to a multicast router on a local interface, using a multicast group membership discovery protocol such as IGMP or MLD.
-Multicast traffic reaches all of the recipients that have joined the multicast group
-Multicast traffic does not reach networks that do not have any such recipients (unless the network is a transit network on the way to other recipients)
-The number of identical copies of the same data flowing over the same link is minimized.
To satisfy these requirements, multicast routing protocols calculate a multicast distribution tree of recipients.
2.3 How to Define Cost of the link
There are various factors, on the basis of which, cost of a link is defined. These factors include Bandwidth, delay, throughput (no. of packets lost) etc. If a link has high cost, it means, it has less efficiency and will cause more packets to be lost. So, our approach is to use kruskal’s algorithm, which choose the links of least-cost, to transmit the packet from sender to receiver(s).
2.4 Minimum Spanning Tree
A tree is a connected (undirected) graph with no cycles.
Given an undirected and connected graph G = (V, E), a spanning tree of G is a sub graph
illustration not visible in this excerpt
If the graph G is weighted, then a minimum spanning tree (MST) of G has the smallest edge-weight sum among all spanning trees of G. Note that when all the edges of the graph have distinct weights, the MST is unique. If V = {v0, v1... vi, vj}, then the MST has n - 1 edges.
These edges must be chosen among potentially n (n - 1)/2 candidates. This gives a lower bound
on the number of operations required to compute the MST since each edge must be examined at
least once.
For convenience, we henceforth refer to the weight of edge (vi, vj) as the distance separating vi and vj and denote it by dist(vi, vj).We are using this approach , because, it helps in reduction the traffic. Since, the packet will follow only the spanning path.
Kruskal’s Algorithm (G, w)
Step 1. A[Abbildung in dieser Leseprobe nicht enthalten]
Step 2. for each vertex v€V [G]
Step 3. do MAKE-SET(v)
Step 4. Sort the edges of E into non-decreasing order by weight ‘w’
Step 5. for each edge (u,v)€ E , taken in non-decreasing order by weight. Step 6. do if FIND-SET (u) FIND-SET(v)
Step 7. then A AU{(u,v)}
Step 8. Union (u, v)
2.5 Ack-Implosion Problem
For a reliable multicasting, every receiver send acknowledgements back to the sender and it is difficult to provide scalability to large numbers of receivers. In this way, for large number of receivers, there could be lack of acknowledgement-packets returning to the sender for every packet that it transmits. Although, using a specialized protocol that very fewer acknowledgements than TCP can still cause an ACK implosion whenever a packet is lost during transmission because every receiver would notify the sender of the missing packet.
illustration not visible in this excerpt
Figure 2.2 Ack-implosion Problem
2.6 IGMP
The following describes the basic operation of IGMP, common to all versions. Note that a multicast router acts as both an IGMP host and an IGMP router in this and following descriptions, and as a result can respond to its own IGMP messages.
-If a host wishes to join a new multicast group, it sends an unsolicited IGMP Report message for that group.
-A local router picks up the IGMP Report message and uses a multicast routing protocol to join the multicast group.
-Periodically, a special router called the Querier broadcasts IGMP Query messages onto the LAN to check which groups the local hosts are subscribed to.
-Hosts respond to the Query messages by sending IGMP Report messages indicating their group memberships.
All routers on the LAN receive the Report messages and note the memberships of hosts on the
LAN. If a router does not receive a Report message for a particular group for a period of time, the router assumes there are no more members of the group on the LAN, and removes itself from the multicast group.
Note that all IGMP messages are raw IP datagram, and are sent to multicast group addresses, with a TTL of 1. Since raw IP does not provide reliable transport, some messages are sent multiple times to aid reliability.
Sending Group Membership Queries
Only one router sends IGMP Query messages onto a particular LAN. This router is called the Querier. IGMPv1 depended on the multicast routing protocol to decide which router was the Querier. IGMPv2 introduced a Querier election process, which works as follows.
By default, a router takes the role of Querier. If a Querier receives an IGMP Query message from a router on the same interface and with a lower IP address, it stops being the Querier. If a router has stopped being the Querier, but does not receive an IGMP Query message within a configured interval, it becomes the Querier again.
Responding to Group Membership Queries
Ordinary LAN routers typically forward multicast traffic onto all other LAN segments. Therefore, the Querier does not need to know exactly which hosts on the LAN require data for a particular multicast group. It only needs to know that one host requires the multicast data.
To avoid a ‘storm’ of responses to an IGMP Query message, each host that receives this message starts a randomized timer for each group that it is a member of. When this timer pops, the host sends an IGMP Report message, which is addressed to that group. Any other hosts that are members of the group also receive the message, at which point they cancel their timer for the group. This mechanism ensures that at most one IGMP Report message is sent for each multicast group in response to a single Query.
Improving Group Membership Latency
IGMPv2 introduced a Leave Group message, which is sent by a host when it leaves a multicast group for which it was the last host to send an IGMP Report message. Receipt of this message causes the Querier possibly to reduce the remaining lifetime of its state for the group, and to send a group-specific IGMP Query message to the multicast group.
Note that the Leave Group message in not used with IGMPv3, as its source address filtering mechanism provides the same functionality.
2.7 Flooding Vs. Broadcasting
Flooding is the first strategy that comes to our mind in the reference of multicasting. In this technique, a router receives a packet and, without even looking at the destination group address, sends it out from every interface except the one from which it was received. Flooding accomplishes the first goal of multicasting, i.e. “Every Network with active members receives the packet”
But the problem with Flooding is: it creates loops. A packet that has left the router may come back again from another interface or the same interface and forwarded again. So, one solution to this problem is, to keep, a copy of the packet for a while and discards any duplicates to prevent loops formation.
Flooding algorithm is guaranteed to find and utilize the shortest path for sending packets because it naturally uses every path in the network. There are no complexities in this routing algorithm; it is very easy to implement.
Of course, there are few disadvantages of the flooding algorithm as well. Because packets are sent through every outgoing link, the bandwidth is obviously wasted. This means flooding can actually degrade the reliability of a computer network.
Unless necessary precautions like hop count or time to live are taken, duplicate copies can circulate within the network without stopping. One of the possible precautions is to ask nodes to track each packet passing through it and make sure that a packet goes through it only once.
Another precaution is called selective flooding. In Selective flooding, nodes may forward packets only in the (approximately) correct direction.
illustration not visible in this excerpt
Figure 2.3 Flooding on the network
In Broadcast communication, One-to-all relationship exists between the source and the destinations. It means, there is only one source host but all others are destination hosts. Broadcasting is a method used in computer networking, which makes sure that every device in the network will receive a (broadcasted) packet. The Internet does not explicitly support broadcasting because of the huge amount of traffic that, it would create and because of the bandwidth that, it would need.
illustration not visible in this excerpt
Figure 2.4 Broadcasting on Network
How Flooding is different from Broadcasting?
Sending a packet to all hosts simultaneously is broadcasting. But flooding does not send packets to all the hosts simultaneously. The packets would ultimately reach all nodes in the network due to flooding. Flooding may send the same packet along the same link multiple times, but broadcasting sends a packet along a link at most once. Several copies of the same packet may reach nodes in flooding, while broadcasting does not cause that problem. Unlike flooding, broadcasting is done by specifying a special broadcast address on packets.
2.8 Various Approaches to Reliable Multicasting
Any reliable multicast protocol requires some recovery mechanism. A generic description of a recovery mechanism consists of a prioritized list of recovery servers/receivers (clients), hierarchically and/or geographically and/or randomly organized. Recovery requests are sent to the recovery clients on the list one-by-one until the recovery effort is successful. There are many recovery strategies available in literature fitting the generic description.
2.8.1 SRM
Scalable Reliable Multicast (SRM) is a simple and robust retransmission-based protocol.SRM uses IP multicast to multicast messages to all the members of the reliable multicast group. In turn, IP multicast uses underlying spanning trees to disseminate these messages to all group members in a best-effort manner, i.e., with no delivery or performance guarantees. Packet recovery in SRM is initiated when a receiver detects a loss and schedules the transmission of a request; an error control message requesting the retransmission of the missing packet. If a request for the same packet is received prior to the transmission of this local request, then the local request is rescheduled by performing an exponential back-off. When a group member receives a request for a packet that it has already received, the group member schedules a reply; a retransmission of the requested packet. If a reply for the same packet is received prior to the transmission of this local reply, then the local reply is cancelled. Using this scheme, all session members participate in the packet recovery process and share the associated overhead. SRM minimizes duplicate error control and retransmission traffic through deterministic and probabilistic suppression. These suppression techniques pres transmitted for each loss. Deterministic suppression prescribes that request and replycribe how requests and replies should be scheduled so that only few requests and replies are scheduling timers be set proportionately to the distance from the source and the requestor, respectively. Thus, the requests of ancestors suppress those of their descendants. Probabilistic suppression prescribes that members that are equidistant from the source and the requestor probabilistically vary the scheduling times of their requests and replies, respectively. Thus, sibling requestor and replier hosts are afforded the opportunity to suppress each other. Unfortunately, suppression introduces a trade-off between the number of duplicate requests and replies and the recovery latency — the scheduling of requests and replies must be delayed sufficiently so as to minimize the number of duplicate requests and replies.
2.8.2 RMTP
RMTP is based on a hierarchical structure in which receivers are grouped into local regions or domains and in each domain there is a special receiver called a designated receiver (DR), which is, responsible for sending acknowledgments periodically to the sender, for processing acknowledgment from receivers in its domain, and for retransmitting lost packets to the corresponding receivers. Since lost packets are recovered by local retransmissions as opposed to retransmissions from the original sender, end-to-end latency is significant reduced, and the overall throughput is improved as well. Also, since only the DR’s send their acknowledgments to the sender, instead of all receivers sending their acknowledgments to the sender, a single acknowledgment is generated per local region and this prevents acknowledgment implosion. Receivers in RMTP send their acknowledgments to the DR’s periodically, thereby simplifying error recovery. In addition, lost packets are recovered by selective repeat retransmissions, leading to improved throughput at the cost of minimal additional buffering at the receivers.
2.8.3 Light-weight Reliable Multicast Protocol
LRMP provides a minimum set of functions for end-to-end reliable multicast network transport suitable for bulk data transfer to multiple receivers. LRMP is designed to work in heterogeneous network environments and support multiple data senders. A totally distributed control scheme is adopted for local error recovery so that no prior configuration and no router support are required. Subgroups are formed implicitly and have no group leaders. Packet loss is reported upon a random timeout first to the lowest level subgroup, then to a higher subgroup and so on until it is repaired. This simple scheme is rather efficient in duplicate NACK and repair suppression.
CHAPTER - 3 SYSTEM DESIGN
3.1 Design Objectives
The design of RSM2 is motivated by the local recovery scheme of RMTP and other existing reliable multicast protocols. RSM2 is intended to take advantages from the existing protocols and avoid their drawbacks. The model has been designed in order to meet the top level goals which must be met at very first stage. Since, many of reliable multicast issues are still under research; some second level goals are established and are expected to be fully met at the second stage.
In the Internet environment, hosts are generally interconnected via heterogeneous links and dispersed world-wide. The quality of service of their network links such as the bandwidth and packet propagation time varies from one to another. The primary goal is thus to provide a reliable data delivery service over such a heterogeneous environment.The goals of RSM2 are summarized as follows:
- It must ensure reliable data delivery.
- It must scale well to large group of users.
- It must continue to work for majority of receivers, even in case where some receivers may leave their group, or some new receivers join the group, during the transmission.
- It must provide reasonable performance.
- It must work in Infrastructured wireless networks.
[...]
-
-
-
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X.