The increasing number of Internet users in the last

decades led to the use of devices and applications with

a huge amount of network resources in order to satisfy end

users requirements. However, the performance of the

traditional networks types which may be wired or wireless

is not satisfactory enough: the cost is very high

or the

throughput is not sufficient. Wireless Mesh Network

(WMN)

is one of new emerging key technology in wireless network

that provide adaptive, flexible and cost effective structural

network [12]. The WMN is built on routers that works by

transferring packets from one node to destination node that

is out of range of sender node. Due to flexible and adaptive

infrastructure, the routers act as multi-hope access point to

the internet for mesh clients’ nodes. These mesh client

nodes also connect with the other nodes to make whole

network adaptive. Various devices such as mobiles, laptops

and PDAs can access the network through WMN at anytime

and anywhere. The adaptive and flexible nature of WMN

makes its integration in any network easy, such as cellular,

wireless, WiMax and WiFi networks [13]. Due to large

number of nodes involvement in

communication routing is the most important issue in this

network. Routing is a mechanism through which the packet

can transfer from source to ultimate destination. Due to

self-configured and self-awareness features of WMNs, it is

expected that in WMN the nodes can decide best path

automatically. Efficient communication in WMN depends

on these routing decisions.

In this study, we discuss the contribution of the

combination of sender-initiated and receiver-initiated

classes in solving the recovery latency problem involving

routers capable of executing personalized services

according to the received packets. To do this, we present

a comparative study between two reliable multicast

protocols: Protocol DyRAM (Active Dynamic Replier

Reliable Multicast) [8], which represents the protocols of

the class receiver-initiated, and the protocol AMRHy

(Active Multicast Reliable Hybrid) [1] that combines both

sender-initiated classes and receiver-initiated. Our study

will show the impact of the depth of the multicast tree, the

group size and the probability of loss on the performance

of both protocols. The remainder of this paper is organized

as follows: related works are discussed in Section 2.

Section 3 describes the network model and basic

definitions and assumptions. Section 4 describes the

behavior of the two analyzed protocols giving the recovery

diagram of each entity. Section 5 presents the results of the

analytical analysis, and the last section concludes this paper

and sets directions for future works.

The comparative study between multicast protocols

have been studied in many context, the first comparative

analysis between the sender-initiated and receiver-initiated

classes of reliable multicast protocols was made by Pingali

et al [9]. This analysis showed that the protocols of the

receiver-initiated class are more scalable than sender-

initiated class because the maximum throughput of sender-

initiated class depends on the number of receivers, whereas

it is not the case in the receiver-initiated class. Levine et al

[5] have extended this work to that the organization of all

Delay of Reliable Multicast Protocols in Wireless Networks

ASMA BENMOHAMMED

Laboratoire MISC, Université Abdelhamid Mehri- Constantine 2,

25000 Constantine, ALGERIA

MERNIZ SALAH

Laboratoire MISC, Université Abdelhamid Mehri- Constantine 2,

25000 Constantine, ALGERIA

Abstract: Wireless Mesh Network (WMN) plays important roles towards the next generation wireless

networking. It is a key technology to support wireless multi-hop networks. Due to dynamic routing nature of

WMNs, the optimization of routing protocol is most critical task. Our work consists on the study and the analysis

of the performances of two reliable multicast protocols based on active networks: AMRHy(Active Dynamic

Replier Reliable Multicast) and DyRAM(Active Multicast Reliable Hybrid). This analysis will allow us to show

the contribution of the combination of the class receiver-initiated and sender- initiated in solving the reliability

problem involving the active routers.

Keywords: reliability, Active networks, Sender- initiated, Receiver-initiated, DyRAM, AMRHy, loss recovery,

Delivery time

Received: June 12, 2022. Revised: July 13, 2023. Accepted: August 15, 2023. Published: September 18, 2023.

1. Introduction

2. State-of-the-art

WSEAS TRANSACTIONS on COMMUNICATIONS

DOI: 10.37394/23204.2023.22.9

Asma Benmohammed, Merniz Salah

E-ISSN: 2224-2864

104

Volume 22, 2023

the receivers in a hierarchical structure ensures scalability

and enhance performance. They also showed that the

protocols using the receiver-initiated class cannot

guarantee the reliability in an environment with limited

caches. Another comparative analysis of sender-initiated

and receiver-initiated classes was presented by Maihöfer

and Rothermel [7]. Their analysis showed that the

protocols of the receiver-initiated class achieve better

scalability but those of the sender-initiated class ensures

reduced latencies. On the other hand, the efficiency of

bandwidth has been subject of several analytical studies.

The analysis of generic reliable multicast protocols was

made by Kasera et al [4] and has shown that local recovery

approaches provide a significant performance of

bandwidth and delay consumption. In [6], Maihöfer

presented an analytical evaluation of the bandwidth of

generic reliable multicast protocols and has shown that

hierarchical approaches provide not only high throughput

but also consume less bandwidth. This work was extended

by comparing the class receiver-initiated and the

combination of classes the sender-initiated and receiver-

initiated by Derdouri et al [2]. The study focused on the

bandwidth consumption and the throughput and showed

that the combination of the classes is more scalable and

consumes less bandwidth than receiver-initiated class

especially when the network is unreliable.

In this paper we extend this comparison to the

recovery latency considering an unreliable backbone where

the data packets losses often occurs within this

backbone,

unlike the previous analysis made on the reliable

multicast

that considers the backbone as being reliable.

WMN is a dynamic self-organized, self-configured

and self-maintained, multi hoped packet network. It

consists of number of nodes that are connected through

wireless media and arranged in a mesh topology. These

nodes can automatically link and leave the network at

anytime.WMN provides services at anywhere and anytime

even if no fixed infrastructure exists at that place. The

nodes in WMN can be act as a both router and a host,

but generally it is categories as a two types of nodes: Mesh

clients (MCs) and Mesh routers (MRs). MRs are fixed and

build the backbone infrastructure of the network where

MCs are usually mobile and roam among these MRs. These

MRs are gateway to internet where MCs can connect to the

MRs and other MCs also. The fixed backbone

infrastructure of network provides multi- hopping access

services to the internet for MCs. The route for packet

communication is selected by using certain routing

protocols. Mesh solutions have many advantages over

traditional wireless networks such as low costs, easy to

maintain, large scale deployment, robustness and greater

coverage area .

Fig. 1. Example of a mesh network

The network model used for the evaluation of reliable

multicast protocols consists on constructing a multicast tree

through a wireless mesh network. The root of this tree

represents the bridge that connects the Mesh network to the

Internet; the leaves of this tree are the Mesh Clients. The

intermediate nodes of the tree represent the Mesh Routers

that are located at different levels to the source. In the

context of wireless mesh networks, all routers in the

multicast tree are considered of being active and can

perform customized treatments on packets passing through

them (data packets or acknowledgment (NAKs and

ACKs)).

The first active service supported by active routers is

the data packets cache for a fixed period to ensure recovery

of lost data packets locally. The second active service

supported by the active routers is the aggregation and

suppression of identical NAKs and ACKs. The third active

service consists in the subcast functionality, where repair

packets are sent only to the affected receivers avoiding the

problem of receptors exposure. The fourth active service is

the dynamic election of a replier providing a local loss

recovery and ensuring a load balanced on the subgroup. To

assess the impact of the combination of the two classes, it

is considered that the two protocols AMRHy and DyRAM

benefit of all active services.

For the delay analysis, we consider a network model

with multi-level multicast tree. A source diffuses data

packets through the tree to R receivers distributed

according to the topology N=R/B, the receptors are divided

into subgroups of B receptors (R1, R2, …, RB) connected

to the source through an active router As, wireless link that

connects the source to the active router is called source-

link. Similarly, the wireless link connecting the active

router At to each of the receivers is called tail-link (see

Figure 2). The wireless links connecting the active routers

between them are the backbone links. We consider that the

source links, terminals and those of a Backbone

respectively have a loss probability Pl.. Therefore, the

probability of end to end perceived by a receiver is P =

1 − (1 − Pl)h where h is the depth of multicast tree. Unlike

the analysis made on reliable multicast, we assume that the

backbone is not reliable and that data packets can be lost in

the Backbone. We suppose initially, that the NAKs (ACKs)

3. Environment and Network Model

3.1 Environment

3.2 Network Model

WSEAS TRANSACTIONS on COMMUNICATIONS

DOI: 10.37394/23204.2023.22.9

Asma Benmohammed, Merniz Salah

E-ISSN: 2224-2864

105

Volume 22, 2023

𝒓

𝐫𝐞𝐜𝐨𝐮𝐯 𝐬

𝑫

𝐫𝐞𝐜𝐨𝐮𝐯 𝐫

𝒓

are never lost and they follow the same path traced by the

data packet in order to benefit from active service.

Fig. 2. Network Model

AMRHy and DyRAM are two protocols that use

active services in routers. Each one adopts a different

strategy to solve the problems of scalability. DyRAM is

based on receiver-initiated class in which the responsibility

The figures (figure 4 and figure 5) show respectively the

recovery policy used by source to recover a lost data packet

and the dynamic election of a replier by the active router.

Fig. 4. DyRAM Source recovery

The recovery of the lost packet is done by the source is

calculated as the following:

𝑬[𝚫𝐃 ] = 𝐍𝐒𝐓 + 𝑫𝑻𝑫 + 𝑬[𝒏𝒑𝒅𝑫] + 𝑬[𝒅𝒑𝒅𝑫] +

𝑬[𝑾𝑫] + 𝑬[𝑾𝑫] + 𝟐𝑬[𝑿] + (𝑻𝒐𝒖𝒕𝒓+𝑬[𝑾𝑫]+𝑬[𝒀])𝒑

of loss detection is attributed to the receivers. However, the

responsibility of loss detection in AMRHy is distributed

𝒔 𝒓 (𝟏−𝒑)

(1)

between the source and receivers by combining the sender-

initiated and receiver-initiated classes. In this hybrid

approach, the source supports the losses that occur in the

source link while receivers care for those that occur in the

terminal links.

Fig. 3. DyRAM and AMRHy in protocol classification

Where 𝑫𝑻𝑫 is the time for active router to elect a

replier , NST is the NACK Suppression Timer,

[𝒏𝒑𝒅𝑫] 𝑖𝑠 the average is required time for a NAK to be

received by the source, 𝑬[𝒅𝒑𝒅𝑫] is the average required

time for a data packet sent by the source to be received by

any receiver. 𝑬 [𝑾

]

is the average waiting time for a data

𝒓

packet to be arrived to destination , 𝑬[𝑿] is the load at the

receiver , 𝒑𝒋−𝟏(𝟏 − 𝒑) is the probability of (j-1)

retransmission of the packet until it is correctly received by

any receiver . 𝑻𝒐𝒖𝒕𝒓is the timer at the receiver.

The recovery of the lost packet is done by a dynamic

elected replier is calculated as the following:

𝑬[𝚫𝐃 ] = 𝐍𝐒𝐓 + 𝑫𝑻𝑫 + 𝑬[𝒏𝒑𝒅𝑫] +

𝑬[𝒅𝒑𝒅𝑫] + 𝑬[𝑾𝑫] + 𝟐𝑬[𝑿] + (𝑻𝒐𝒖𝒕𝒓 + 𝑬[𝑾𝑫] +

DyRAM is an active reliable multicast protocol,

based on the hierarchical approach. It adopts a local

recovery scheme based on receiver-initiated class; the

receivers are responsible of the losses detection, and

possibly the retransmission of the lost data packets. This

protocol is distinguished from other reliable multicast

protocols by its innovative features:

•

For each lost packet, a dynamic election of a replier is

performed. It is elected from among the receivers that have

correctly received the data packet.

•

The Emulation of positive acknowledgments due to the

addition of new fields in the headers of the control packet

of the protocol.

•

The Subcast repair packets only to receivers who actually

lost the data packets.

[𝒀])𝒑/(𝟏 − 𝒑) (2)

Fig. 5. DyRAM dynamic election of a replier

4. Protocol Description

4.1 Protocol DyRAM

WSEAS TRANSACTIONS on COMMUNICATIONS

DOI: 10.37394/23204.2023.22.9

Asma Benmohammed, Merniz Salah

E-ISSN: 2224-2864

106

Volume 22, 2023

𝐫𝐞𝐜𝐨𝐮𝐯 𝐬 𝒔

AMRHy overcomes the inconveniences of the

"receiver-initiated" class by combining the two classes

The recovery of the lost packet is done by the

receiver elected as a replier is calculated as the following:

𝑬[𝚫𝐀 ] = 𝑬[𝒏𝒑𝒅𝑨] + 𝑬[𝒅𝒑𝒅𝑨] + 𝑬[𝑾𝑨] +

"sender-initiated" and "receiver-initiated". The

𝐫𝐞𝐜𝐨𝐮𝐯 𝐫

𝑬[𝑾𝑨]

+ 𝟐𝑬[𝑿] + (

𝑻

𝒂

+ 𝑬[𝑾𝑨] + 𝑬[𝒀]) 𝒑⁄(𝟏 − 𝒑)

combination of these two classes has resulted in the

𝒓

coexistence of positive and negative acknowledgments.

The positive acknowledgment is used between the

𝒐𝒖𝒕 𝒓

𝒓

(4)

receivers and the source to confirm the correct reception of

a data packet by at least one receiver. It also allows the

active router:

—

To invite the members of his group, having lost the data

packet to report the loss before the packet is removed

from the cache.

—

To inform the members of the group by the address of

the dynamically replier elected for future repairs without

active services,

—

To inform the members of his group, who received the

data packet correctly, to make a local suppression of

their corresponding ACKs

—

Delete the cache of the data packet in question.

The recovery of the lost packet is done by the

source is calculated as the following:

𝑬[𝚫𝐀 ] = 𝑬[𝒅𝒑𝒅𝑨] + 𝑬[𝒂𝒑𝒅𝑨] + 𝑬[𝑾𝑨] +

𝑬[𝑾𝑨] + 𝟐𝑬[𝑿] + 𝑬[𝑾𝑨] + 𝟐𝑬[𝒁] + (𝑻 ) 𝒑⁄(𝟏 − 𝒑)

In this section, we use the model described before

to analyze numerically the two protocols. The study is done

according to the overall delay calculated for each protocol.

we determine the influence of three parameters: the size of

the local group (B), the depth of the backbone (H) and the

probability of loss (P) on the delay of losses recovery at the

packet level of data. For the numerical evaluation, we adopt

the same measures as those taken in [11]:

Figure 8 show that the protocol AMRHy is more

efficient in terms of overall time than the protocol DyRAM

and this regardless of the increase of the local group B, this

is caused by the combination of the sender-initiated and

receiver-initiated classes adopted by AMRHy allowing a

fair distribution of load between the source, the active

routers and receivers. In addition

𝒂 𝒔

(3)

𝒐𝒖𝒕

On the other hand, it allows the source to release

the transmission buffer associated with the data packet. The

negative acknowledgment is used locally between the

router

and the active receivers of its group to report the loss

of a data

packet.

Fig. 6. AMRHy Source recovery

Fig. 7. AMRHy receiver elected as a replier recovery

Fig. 8. Overall delay of the protocols A and D in function of the

size of the local group (h = 2, p = 0.01)

4.2 Protocol AMRHy

5. Experimental Results

5.1 Influence of the Size of the Local Group

B on the Overall Time

WSEAS TRANSACTIONS on COMMUNICATIONS

DOI: 10.37394/23204.2023.22.9

Asma Benmohammed, Merniz Salah

E-ISSN: 2224-2864

107

Volume 22, 2023

Fig. 9. Overall delay of the protocols A and D in function of the

depth of the Backbone (B=150, p=0.1)

Fig. 9 show that the depth of the backbone affects

the overall delay of both protocols DyRAM and AMRHy,

the protocol AMRHy still more efficient compared to

DyRAM for the same reason mentioned before. The

figures also show that the depth of the backbone affects

overall delays of both protocols AMRHy and DyRAM.

Fig 10: Overall delay of the protocols A and D in function of

the loss probability (B=90, h=6)

Figure 10 show that the protocol AMRHy is more

efficient than the protocol DyRAM in terms of overall

delay thanks to the benefits of the aggregation and

suppression of local ACKs service through a combination

of both hierarchical and based on timers’ approaches.

These figures show that the probability of loss has a major

influence on the overall delay in the protocol

DyRAM

This

is due to the number of NAKs generated as many times as

the packet is lost. The probability of loss also affects the

protocol AMRHy but it is negligible compared to that

protocol DyRAM.

This article was devoted to the comparative study of

two protocols AMRHy and DyRAM with the analytical

analysis. The results obtained show the exemplary

behavior of the protocol AMRHy for environments highly

unreliable transmissions with a dense population.

Indeed, it was observed that unlike the protocol

DyRAM, delivery times in the protocol AMRHy are

subject to a minimal degree of changes when the loss

probability, the depth of the Backbone and the group size

are increased.

Based on the results, one can conclude that the

protocol AMRHy has the potential needed to migrate to a

wireless environment (sensor networks, embedded

systems, IoT, ect…). where the loss rate is too high . This

potential can be summarized as:



Simplicity of treatment



Support for high loss rates



Tolerance of a high density of receptors

As perspective, we will expand our analysis to the

loss of acknowledgment (ACK and NAK) on the

backbone.

5.2 Influence of the Depth of the Backbone H

on the Overall Time

5.3 Influence of the Loss Probability P on the

Overall Time

6. Conclusion

References

[1]. Derdouri. L, Pham. C, Benmohammed. M,

AMRHy: Hybrid Protocol for a Reliable

Multicast Transport in Active Networking

Environments, International Journal on

Computer Science and Network Security

(IJCSNS), vol. 8(10), pp68-78, October 2008.

[2]. Derdouri. L, Pham. C, Benmohammed. M, A

Comparative Analysis of Reliable Multicast

Protocols in Active Networking Environment,

International Review on Computers and

Software (IRECOS), Vol.3(3), Praize Worthy

Prize, May 2008.

[3]. Derdouri Lakhdar, Une Approche Hybride

pour le Transport Multicast Fiable dans un

Environnement Actif, Thèse de Doctorat en

informatique, Université Mentouri

Constantine, Juin 2009

WSEAS TRANSACTIONS on COMMUNICATIONS

DOI: 10.37394/23204.2023.22.9

Asma Benmohammed, Merniz Salah

E-ISSN: 2224-2864

108

Volume 22, 2023

[4]. Fenner. W, Internet Group Management

Protocol, Version 2, Internet standard RFC

2236, November 1997.

[5]. Hinden. R and Deering. S, IP version 6

Addressing Architecture, Internet standard

RFC 2373, July 1998.

[6]. Hussain. S. A, Active and Programmable

Networks for Adaptive Architectures and

Services, Auerbach Publications by Taylor &

Francis Group, LLC, December 2006

[7]. Kasera. S, Kurose. J, Towsley. D, A

comparison of server-based and receiver-based

local recovery approaches for scalable reliable

multicast, in Proceedings of INFO-COM’98,

San Francisco, CA, March 1998.

[8]. Legedza. U, Guttag. J, Using Network-level

Support to Improve Cache Routing, 3rd

International WWW caching Workshop,

Manchester, England, June 1998.

[9]. Maimour. M and Pham. C, A throughput

Analysis of Reliable Multicast Protocols in

Active Networking Environment, in

Proceeding of the sixth IEEE Symposium on

Computers and Communications, 2001

[10]. Oggioni. B, Charpentier. M, Algorithme des

télécommunications: Wirless Mesh

Networking, Juin 2008

[11]. Sneha Kumar Kasera, James F. Kurose,

Donald

[12]. F. Towsley: Scalable Reliable Multicast Using

Multiple Multicast Groups. In measurement

and modeling of Computer System, 1997.

[13]. Liang Zhao, Ahmed Y. Al-Dubai,Routing

Metrics for Wireless Mesh Networks: a survey,

Springer Berlin Heidelberg. 127: 311-316.

[14]. Paramjeet Kaur Bedi, Yadu Nagar, Amit, Rajni

Yadav, Study Of Routing Protocols: Single

And Multipath For WMN”, (TIJCSA),March

2012, Volume 1, No. 1, ISSN No. 2278-1080.

[15]. D. Kandris and all, LEACH-based hierarchical

energy efficient routing in wireless sensor

networks, AEU - International Journal of

Electronics and Communications, Vol. 169,

Sept. 2023.

[16]. Zhibin Liu, Xinshui Wang, Energie-balanced

routing in wireless sensor netword with

reinforcement learning using greedy action

chains, Soft Computing Journal, June 2023.

[17]. Maihöfer et Rotherme, A delay analysis of

generic multicast transport protocols, IEEE

International Conference on Multimedia and

Expo, 2001. ICME 2001. Tokyo Japan.

[18]. Pearl Antil, Amita Malik, An Energy-Efficient

Routing Model for Scale-free Wireless Sensor

Networks, Indian Journal of Engineering &

Materials Sciences Vol. 30, April 2023, pp.

338- 345 DOI: 10.56042/ijems.v30i2.818

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy)

The authors equally contributed in the present

research, at all stages from the formulation of the

problem to the final findings and solution.

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

No funding was received for conducting this study.

Conflict of Interest

The authors have no conflicts of interest to declare

that are relevant to the content of this article.

Creative Commons Attribution License 4.0

(Attribution 4.0 International, CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en

_US

WSEAS TRANSACTIONS on COMMUNICATIONS

DOI: 10.37394/23204.2023.22.9

Asma Benmohammed, Merniz Salah

E-ISSN: 2224-2864

109

Volume 22, 2023