A Cluster-Based Approach for Supporting QoS in ... - Semantic Scholar

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh Department of Computer Science, University of Jordan P.O. Box 13838, Amman, 11942, Jordan {wesmoba, hushaidan, azzam.sleit, mohd.qat}@ju.edu.jo doi:10.4156/jdcta.vol5. issue1.1

Abstract

Providing Quality of Service (QoS) for Mobile Ad Hoc Networks (MANET) is a challenging issue due to th e dynamic nat ure and limit ed res ources of such in frastructureless wireless net works. T his paper proposes a ne w app roach f or supporting QoS in clus tered MANET . T he proposed Cluster-Based Qo S (CBQoS) pro vides MA NET with i nter-cluster/intra-cluster ser vice diff erentiation and aims to improve the overall performance of the network by increasing the overall network throughput and decreasing the overall end-to-end delay. CBQoS can be implemented as a standalone service differentiation system or be used wi th Differentiated Services (DiffServ) or o ther QoS m odels to enha nce Qo S provisionin g i n MANET. The simulation res ults s how t hat the proposed approach achieves significant improvement in MANET's performance and QoS support.

Keywords: Mobile Ad Hoc Network, Quality of Service, Wireless Communications, Clustering 1. Introduction Mobile Ad hoc Networks (MANET) are wireless networks that can be easily deployed when and where needed without the need to a fixed infrastructure or centralized administration. A MANET consists of a collection of wireless devices (nodes) that communicate with each other using shared wireless medium. Each node in the network is required to be capable of forwarding (i.e. acting as a router), in addition to its role in sending and receiving data packets. Potential applications of MANET include mobile conferencing, emergency services, disaster recovery, and battlefield operations [1, 2]. Supporting Quality of Service (QoS) in multimedia networks is desirable to improve the performance of communications and satisfy the requirements of different network applications [3]. Research has been conducted on supporting QoS for the Internet and introduced several QoS models such as Integrated Service (IntServ) [4] and Differentiated Services (DiffServ) [5]. However, it is more difficult and challenging to provide QoS in MANET due to many limitations and constraints imposed by MANET including its infrastructureless nature, dynamic topology, and low communication bandwidth; in addition to the limited capabilities of wireless devices. Previous studies showed that it is not straightforward to adopt pure IntServ or DiffServ in MANET since these models were proposed for relatively high speed wired networks [6]. Many studies have focused on QoS provisioning for MANET and introduced new QoS models like FQMM (Flexible QoS Model for Mobile Ad-Hoc networks) [6], SWAN (stateless network model) [7], and HQMM (Hybrid QoS Model for Mobile Ad-Hoc Networks) [8]. Other studies proposed QoS resource reservation signaling systems such as INSIGNIA [9], and QoS Medium Access Control (MAC) such as IEEE 802.11e [10]. This paper proposes CBQoS which is a cluster-based approach to support QoS provisioning in clustered MANET. The clustering function is not part of this work. A clustering algorithm should be used to partition the network into clusters as a basis for CBQoS implementation. Several clustering algorithms cited in the literature can be used for this purpose such as Link-Cluster Architecture (LCA) [11, 12], Weighted Clustering Algorithm (WCA) [13], and Max-Min D-cluster Algorithm [14]. Further improvement to the clustering algorithm can be gained by adopting the Distributed Clusterhead Architecture [15, 20] which improves the network performance and eliminates the single point-of-failure problem in clustered MANET.

-1-

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011

The rest of this paper is organized as follows. Section 2 proposes a cluster-based QoS approach for Mobile Ad Hoc Network. Section 3 presents the simulation model used to evaluate the proposed approach with analysis of results. Finally, section 4 concludes this paper.

2. A Cluster-Based QoS for MANET Packet-switched networks utilize routers for supporting multi-hop data transmission. Routers receive packets, buffer them in specific forwarding queues, and forward them to the next hop according to a specific scheduling mechanism. Multiple transmission flows may pass through the same router contending for the same buffer space and transmission bandwidth. When QoS is not provided in a specific router, all packets are treated in a First-In-First-Out (FIFO) forwarding mechanism. In high contention periods, the router’s buffer becomes full. Therefore, incoming packets have no place in the forwarding queues. These packets are dropped and must be retransmitted (if their application requires so). In MANET, nodes function as routers to facilitate transmission between nodes that can not directly reach each other. Intermediate nodes for a specific transmission flow act as routers for the packets of this flow. In clustered MANET, nodes are grouped together into clusters to make a hierarchical control environment and facilitate routing. Communication between nodes in clustered MANET can be classified into inter-cluster and intracluster. In Intra-cluster communication (IA), the source and destination nodes both belong to the same cluster. The packets are usually sent from the source node to the clusterhead (CH) which forwards the packet to the destination node. However, in Inter-cluster communication (IE), the source and destination nodes are in different clusters and packets are sent from the source node to its CH which forwards the packet to the corresponding node's cluster. IE Packets may pass through multiple hops and clusters until reaching the corresponding CH which delivers packets to destination.

2.1. The Basic Idea The Idea of CBQoS is to provide a cluster-based service differentiation, where the forwarding modules of intermediate nodes (routers) classify packets into inter-cluster packets (IE) and intra-cluster packets (IA). The router provides better treatment (higher forwarding priority and lower dropping probability) for IE packets than IA packets. The goal of the CBQoS is to improve the overall network performance allowing for better service and broader range of usable applications on MANET. CBQoS improves the network performance by reducing the probability of discarding or dropping inter-cluster packets since dropping a packet that has traveled across many clusters results in degrading the performance than that of dropping an intra-cluster packet which can be retransmitted with lower cost of buffer-space and transmission bandwidth. It gives higher priority to packets that have traveled longer across network clusters. This approach decreases the delay encountered by inter-cluster traffic (which is usually high) and eliminates intra-cluster packets from starving inter-cluster ones. CBQoS provides nodes in MANET with traffic classification and queue management mechanisms as described below.

2.2. Model Architecture The components of the CBQoS approach include Traffic Classifier (classifies packets either to intercluster (IE)) or intra-cluster (IA), Packet Scheduler (schedules packets for transmission providing higher priority to IE packets), and Packet Dropper (uses some dropping policy in case of congestion). These components are illustrated in Figure 1.

-2-

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011

Input Packets

IE Queue

Traffic Classifier Output Packets

IA Queue

Packet Dropper Dropped Packets

Figure 1. Traffic Classification/Scheduling in CBQoS 2.2.1 Traffic Classification Traffic is classified into two classes; namely, inter-cluster (IE) traffic and intra-cluster (IE) traffic, where IE traffic gets higher priority. Packets are given intra-cluster forwarding behaviour in their initiating clusters and inter-cluster forwarding behaviour (higher priority) in other clusters. Traffic classification can be done based on two values: the packet's IP TTL field, and the cluster-range (cluster radius) which is determined by the clustering algorithm and assumed to be known to all nodes in the network. When a packet is generated, it is classified in its source node as an intra-cluster (IA) packet. Once this packet arrives at another node, its IP TTL field is checked to determine how many hops has the packet travelled and based on the TTL value together with the cluster-range value, the packet is classified as inter-cluster or intra-cluster and queued accordingly. 2.2.2. Packet Scheduling After a packet gets classified into either IE or IA, it is queued in the relevant queue. Queue management can be achieved as follows. Two queuing priorities (for inter-cluster and intra-cluster packets) are implemented. Inter-cluster packets are given higher priority than intra-cluster packets. Various queuing approaches such as Priority Queuing [16], Weighted Fair Queuing, or Class-Based Queuing can be adopted to implement the CBQoS. 2.2.3. Packet Dropping Another aspect of the queue management policy is the dropping policy which can be implemented to deal with congestion. The dropping policy can be as simple as dropping input packets when the buffers are full. A complex dropping policy may require running statistics and complex measurements. A good dropping policy for CBQoS is the one that provides inter-cluster (IE) packets with dropping probability without drastically starving IA packets. Random Early Detection (RED) [17] can be used as a dropping policy, where inter-cluster packets having less drop probability than intra-cluster ones. RED is a congestion control mechanism that monitors different queues. Each queue is given a specific average length and dropping probability. When a queue exceeds a specific threshold, RED drops its packets with a certain dropping probability.

2.3. Design Choices CBQoS can be designed as a standalone service provisioning approach as it can also be applied over classical DiffServ [5] without any extra header field in IP packets, and without affecting the concept of the DiffServ model.

-3-

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011 Queue

Class 1

Queue

Class n

Queue

IE Queue

Class 0 Higher priority

Class 0

IA Queue IE Queue

Class 1

IA Queue

IE Queue

Class n

a) Original DiffServ Forwarding

IA Queue

b) DiffServ Forwarding Classes

Figure 2. Implementing CBQoS over DiffServ When CBQoS is implemented over DiffServ, packets are firstly classified into forwarding classes using DiffServ, and then CBQoS is applied within each forwarding class as shown in Figure 2.

3. Performance Evaluation In this section, we evaluate CBQoS on clustered MANET through simulation. The following terminologies are used through the rest of this paper: IE: IntEr-cluster traffic IA: IntrA-cluster traffic Overall Throughput: the overall throughput for all flows in the network Overall Delay: the overall average end-to-end delay encountered by all flows in the network Network Power: a network performance metric that measures the throughput to delay ratio [18].

3.1. Simulation Environment Global Mobile Information System Simulator (Glomosim) [19] was used to simulate two network models. 12 13 5

8

4

3

6

10

14 9

1 7 15

2

11

IE Traffic IA Traffic

Figure 3. A clustered MANET with IE and IA Traffic The scenario shown in Figure 3 was used in our simulation, which represents a clustered MANET that occupies a terrain of (2500*2500) meters and consists of 15 nodes grouped in 3 clusters with nodes 1, 6, and 9 as the three clusterheads (CH). A disjoint clustering method was assumed, with the pair (4, 5) and (8, 10) as the distributed gateways (DG) [12]. Each member node in a cluster can communicate with others via its own clusterhead. Gateway nodes provide connections between neighbouring clusters. For the medium access, the IEEE 802.11 MAC is used and for the physical layer, the two-ray propagation pathless model is used. Each simulation experiment was run considering two cases:  QoS OFF is the traditional case where CBQoS is not use. The simulation was run ten times for each simulation experiment, each with different simulation seed. No traffic differentiation was provided.  QoS ON uses CBQoS. The simulation was also run ten times for each simulation experiment, each with different simulation seeds.

-4-

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011

The traffic used in the simulation is Constant Bit Rate (CBR) with packets of 1000 bytes in size. Different traffic loads on the network were used in the simulation to inspect the effect of differentiating inter/intra-cluster communication in each load. Two types of traffic were simulated:  IE traffic is represented by the flow between nodes (1) and (14) in Figure 3.  IA traffic represented by two IA flows in two different clusters in Figure 3. Node (7) sends to node (12), and node (11) sends to node (13). To study the effect of prioritizing IE traffic over IA traffic on different network loads, five different sending rates of IA traffic were used; namely, 200Kbps, 400Kbps, 640Kbps, 800Kbps, and 1Mbps. Results show the behavior of IE traffic (which was fixed at 200Kbps) for various network loads.

3.2. Simulation Results and Analysis Throughput, Delay, and Network Power are considered as the performance metrics to evaluate how the proposed QoS approach contributes in supporting QoS for MANET. 3.2.1. Throughput

Throughput

Kbps

The first performance metric we used to evaluate the CBQoS approach was throughput. In Figures 4, the throughput of IE traffic is shown before and after using CBQoS. Table 1, which lists the percentage of throughput using CBQoS compared to normal throughput, shows that IE gains higher throughput using CBQoS than the traditional case. This gain is due prioritizing IE traffic over IA traffic such that IE packets are queued on the front and scheduled for transmission before IA packets. The gain in IE throughput starts small (2.35%) when the network is lightly loaded (200kbps) and increases gradually with higher network loads. With lightly loaded networks, IE traffic does not suffer high contention from other traffic. In this case, both IE and IA receive high throughput even without the use of CBQoS. This explains the small increase in IE throughput when using CBQoS. 140 120 100 QoS OFF

80

QoS ON

60 40 20 0 200 Kbps

400 Kbps

640 Kbps

800 Kbps

1Mbps

Network Load

Figure 4. Throughput of IE traffic with/without CBQoS IE traffic suffers more as the network load increases because the network becomes congested and the contention increases. This behavior is expected when we are not using CBQoS. With CBQoS, however, IE gets relatively higher throughput but it suffers due to the nature of the Distributed Coordination Function (DCF) of IEEE 802.11 MAC, which gives equal priority for each node in the network. Furthermore, when IE is prioritized on the network layer and scheduled first, it will have the same contention priority to use the wireless medium and since IA flows in neighboring nodes are high, these nodes ask more frequently for medium access causing the node (through which IE passes) to backoff and wait longer before getting a chance to access the medium.

-5-

Kbps

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011 700 600

Throughput

500 400

QoS OFF

300

QoS ON

200 100 0 200 Kbps

400 Kbps

640 Kbps

800 Kbps

1Mbps

Network Load

Figure 5. Throughput of IA traffic with/without CBQoS

Overall Throughput

Kbps

Figure 5 shows the throughput of IA traffic with and without using CBQoS. An increase (of nearly 2% on average) is also gained by IA traffic when using CBQoS, although IE traffic is getting priority over it. This increase is also due to IEEE802.11 MAC nature. Without QoS, IE packets encounter higher contention, waiting longer in the queue and occupying a valuable queuing space which can be rather left to IA traffic, and after queuing in several intermediate nodes, an IE packet gets equal dropping probability as that of IA packets. This waste of queue capacity and wireless medium affects both IA and IE traffics when no cluster-based differentiation is used. 660 640 620 600 QoS OFF 580

QoS ON

560 540 520 500 200 Kbps

400 Kbps

640 Kbps

800 Kbps

1Mbps

Network Load

Figure 6. Overall Throughput with/without CBQoS Table 1. Network Throughput with CBQoS Network IE IA Overall Load Throughput Throughput Throughput 200 Kbps 400 Kbps 640 Kbps 800 Kbps 1Mbps Average

+2.35% +9.64% +17.28% +21.06% +17.53% +13.57%

+0.68% +3.14% +1.64% +2.19% +2.25% +1.98%

+1.09% +3.42% +1.90% +2.50% +2.50% +2.28%

The overall network throughput is depicted in Figure 6, which shows an increase in network throughput by 2.28% when using CBQoS. 3.2.2. Delay The delay is an important performance metric especially for real time applications. Figure 7 shows a significant decrease by 38.75% of the average end-to-end delay for IE traffic when using CBQoS. This is because IE is queued and scheduled before IA traffic. Consequently, the intermediate queuing delay of IE packets is reduced. Delay for IE traffic is small when the network load is light even without QoS support.

-6-

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011

Average end-to-end Delay

2 1.8 1.6 1.4 1.2

QoS OFF

1

QoS ON

0.8 0.6 0.4 0.2 0 200 Kbps

400 Kbps

640 Kbps

800 Kbps

1Mbps

Network Load

Figure 7. IE end-2-end Delay with/without CBQoS As the network load becomes higher, IE traffic suffers higher delay when it is not supported with QoS. With CBQoS support, IE gets very smaller delay in highly loaded network. This explains the behavior of IE delay depicted in Figure 7. Intra-cluster end-to-end delay is shown in Figure 8. We note that the delay of IA packets with and without using cluster-based QoS is almost the same. The delay of IA is a bit higher by 1.63% when using cluster-based QoS as displayed in Table 2. This is expected due to prioritizing IE over IA packets in the cluster-based QoS approach. However, the overall delay encountered by all flows (IE+IA) in the network, is smaller by 22% on average for various network loads when using cluster-based QoS.

Average end-to-end Delay

1.8 1.6 1.4 1.2 QoS OFF

1

QoS ON

0.8 0.6 0.4 0.2 0 200 Kbps

400 Kbps

640 Kbps

800 Kbps

1Mbps

Network Load

Figure 8. IA end-2-end Delay with/without CBQoS The overall end-to-end delay is shown in Figure 9. Table 2 lists the percentage of decrease in end-toend delay using cluster-based QoS compared to the normal end-to-end delay. O verall Average end-2-end Delay

4 3.5 3 2.5 QoS OFF

2

QoS ON

1.5 1 0.5 0 200 Kbps

400 Kbps

640 Kbps

800 Kbps

1Mbps

Network Load

Figure 9. Overall end-2-end Delay with/without CBQoS

-7-

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011

Table 2. Network Delay with CBQoS Network IE IA Overall Load Delay Delay Delay 200 Kbps 400 Kbps 640 Kbps 800 Kbps 1Mbps Average

-3.41% -26.43% -54.20% -54.95% -54.79% -38.75%

+13.50% +2.03% -0.93% -3.21% -3.23% +1.63%

-0.16% -16.59% -29.72% -30.71% -30.58% -21.55%

3.2.3. Network Power The Network Power is a network performance metric which reflects the overall performance of the network. Network power is computed using the formula power = (throughput ^ alpha)/delay [18], where alpha is chosen based on the relative importance of throughput versus delay. An alpha value equal to 1 is used when throughput and delay are equally important. If delay is more important than throughput, alpha should be chosen smaller than one. We use the ratio of the overall network throughput to the overall network delay to represent the Network Power. Choosing alpha=1, and assuming that throughput and delay are equally important and we concluded that cluster-based QoS empowers the network with an average increase of 33.13% as displayed in Table 4, and Figure 10. Results according to Tables 1 and 2 show that cluster-based QoS provides higher network power when delay is of higher importance than throughput since cluster-based QoS contributes to the reduction of delay more than its contribution to the increase in throughput. This conclusion makes it more suitable to apply cluster-based QoS on MANET that are deployed for real time applications such as voice communication. Table 3. Network Power with CBQoS Network Network Load Power 200 Kbps 400 Kbps 640 Kbps 800 Kbps 1Mbps Average

+1.25% +24.00% +44.97% +47.83% +47.60% +33.13%

Table 3 shows that the proposed cluster-based QoS approach achieves significant performance improvement in clustered MANET. This gain in performance is important to support QoS provisioning.

4. Conclusion This paper proposed a cluster-based QoS approach which provides inter-cluster/intra-cluster traffic differentiation for Mobile Ad Hoc Networks. The proposed approach provides a special per-class service differentiation that decreases the overall delay and increases the overall throughput in clustered MANET. Simulation results show that the proposed QoS approach improves the capacity of MANET with higher throughput and lower end-to-end delay. The proposed approach was implemented on the network layer over the classical IEEE 802.11 MAC, which provides equal contention opportunities for all nodes regardless of their status. This approach achieves improvement in MANET performance with cooperation from the MAC layer. The idea of cluster-based service differentiation can be further investigated on other network technologies that use clustering.

-8-

A Cluster-Based Approach for Supporting QoS in Mobile Ad Hoc Networks Wesam Almobaideen, Khaled Hushaidan, Azzam Sleit, Mohammad Qatawneh International Journal of Digital Content Technology and its Applications. Volume 5, Number 1, January 2011

5. References [1] I.F. Akyildiz, T. Melodia, and K.R. Chowdhury, “A Survey on Wireless Multimedia Sensor Networks”, Computer Networks Journal, Vol. 51, No. 4, pp. 951-960, March 2007. [2] I.F. Akyildiz, and E.P. Stuntebeck, “Wireless Underground Sensor Networks: Research Challenges”, Ad Hoc Networks Journal, Vol. 4, No. 6, pp. 669-686, November 2006. [3] Z. Wang, Internet QoS: Architectures and Mechanisms for Quality of Service, Morgan Kuffmann publishers, 2001. [4] R. Braden, D. Clark, and S. Shenker, “Integration in the Internet Architecture: An Overview”, IETF RFC 1633, June, 1994. [5] S. Blake, D. Black, M. Carlson, E. Davies, Z. Wang, and W. Weiss, “An Architecture for Differentiated Services”, IETF RFC 2475, December, 1998. [6] H. Xiao, W.K. Seah, A. Lo, and K.V. Chua, “A Flexible Quality of Service Model for Mobile AdHoc Networks”, IEEE VTC2000-spring, Tokyo, Japan, May, 2000. [7] Rongjie Liu, Guangsheng Cao, Jie Zhang, Pingjian Song, Binge Cui, "Research on Dynamic Web Services Management Based on QoS", JDCTA: International Journal of Digital Content Technology and its Applications, Vol. 4, No. 5, pp. 55-61, 2010. [8] Y. He and H. Abdel-wahab, “HQMM: A Hybrid QoS Model for Mobile Ad-hoc Networks”, In proceedings of the 11th IEEE symposium on Computers and Communications (ISCC'06), 2006. [9] S.B. Lee and A.T. Campbell, “INSIGNIA: In-band Signaling Support for QOS in Mobile Ad Hoc Networks”, In proceedings of the5th International Workshop on Mobile Multimedia Communications (MoMuC, 98), Berlin, Germany, October, 1998. [10] IEEE Standards Board. Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications. Amendment 8: Medium Access Control (MAC) Quality of Service Enhancements, IEEE Standard 802.11e™, 2005 [11] D.J. Baker and A. Phremides, “A distributed algorithm for organizing mobile radio telecommunication networks”, IEEE 2nd international conference on distributed computing systems, Paris, France, pp. 476-483, 1981. [12] M. Gerla and J.T. Tsai, “Multicluster, mobile, multimedia radio network”, ACM-Baltzer Journal of Wireless Networks, Vol.1 No. 3, pp. 255-265, 1995. [13] M. Chatterjee, S.K. Das, and D. Turgut, “WCA: A Weighted Clustering Algorithm for Mobile Ad hoc Networks”, Journal of Cluster Computing, Special Issue on Mobile Ad hoc Networks, Vol. 5, No. 2, pp. 193-204, 2002. [14] Huang Zhi, Liu San-Yang, Qi Xiao-Gang, "Overview of Routing in Dynamic Wireless Sensor Networks", JDCTA: International Journal of Digital Content Technology and its Applications, Vol. 4, No. 4, pp. 19-26, 2010. [15] E. Qaddoura, W. AlMobaideen, and A. Omari, “Distributed Clusterhead Architecture for Mobile Ad Hoc Networks”, Journal of Computer Science, Vol.2, No.7, pp. 583-588, 2006. [16] C. Semeria, “Supporting Differentiated Service Classes: Queuing Scheduling Disciplines”, (white paper), Juniper Networks Inc. Sunnyvale, USA, 2001. [17] S. Floyd, V. Jacobson, “Random Early Detection gateways for Congestion Avoidance”, IEEE/ACM Transactions on Networking, Vol.1, No.4, pp. 397-413, 1993. [18] A. Mankin, K. Ramakrishnan, Gateway Congestion Control Survey, IETF RFC 1254, August 1991. [19] L. Bajaj, M. Takai, R. Ahuja, K. Tag, R. Bagorodia, and M. Gerla. GloMoSim:A Scalable Network Simulation Environment, Technical Report 990027, University of California, 1999. [20] J. Yick, B. Mukherjee, D. Ghosal, “Wireless sensor network survey”, Computer Networks Journal, vol. 52, No. 12, pp. 2292-2330, 2008.

-9-