Sage Journals: Discover world-class research

Abstract

Existing routing protocols for hybrid wireless mesh network neglect the negative impact on network lifetime and route stability caused by energy constraints and mobility of mesh clients. To solve this issue, a regional energy- and mobility-aware routing protocol is proposed in this article. Both energy and mobility features in hybrid wireless mesh network are considered in this routing protocol. On one hand, within the communication transmission range, the intensity and degree of dispersion in energy consumption can be perceived. The node whose neighbor area has sufficient energy can be selected for network services, which can balance energy and extend the network lifetime. On the other hand, the mobility of clients is considered to enhance the stability and reduce network overhead. Therefore, regional energy- and mobility-aware routing protocol can balance network energy consumption, prolong network lifetime, and improve route stability. Simulation results via NS2 show that regional energy- and mobility-aware routing protocol can achieve better performance significantly.

Keywords

Hybrid wireless mesh network routing protocol routing metric regional energy aware mobility aware

Introduction

Due to the features of low cost, highly robust, and high reliability, wireless mesh network (WMN) is one of the key technologies in the next generation of wireless network. The self-organizing and self-healing peculiarities can dramatically reduce the complexity in network deployment and maintenance. WMN can be widely used in many circumstances, such as company network, medical system, and broadband family network. In addition, because of the advantage of seamless broadband connection to the Internet, WMN has attracted increasing concentration from researchers.^1–3

WMN is composed of infrastructure WMN, client WMN, and hybrid WMN. Infrastructure WMN includes static mesh routers with multiple radio interfaces, while client WMN is made up of mobile mesh clients which are always equipped with only one radio interface. Hybrid WMN combines advantages of infrastructure WMN and client WMN, which has stronger network connectivity, higher reliability, and larger coverage.^4,5 Its structure is quite more flexible and closer to actual demand than infrastructure WMN or client WMN.⁶ In hybrid WMN, mesh routers have high capacity and stability, while mesh clients have limited energy and high mobility. Mesh clients can not only communicate with each other by accessing mesh routers but they can also forward packets directly without the help of mesh routers.⁷

Routing protocols aim to provide best routes for two communicating parties, which can greatly influence the network performance. They always centralize in the network layer.⁸ However, current routing protocols mainly focus on infrastructure WMN or client WMN, and the routing protocols designed for hybrid WMN are few. As the multi-hop communication brought by mesh clients causes link diversity and complexity of network condition,⁹ current routing protocols for infrastructure WMN are no longer adaptive for hybrid WMN. Besides, the features of mesh clients can cause the condition that links are unstable and network lifetime is short. Hence, designing a proper routing protocol considering special features of hybrid WMN is quite meaningful. Routing protocols designed for hybrid WMN nowadays have different emphases.¹⁰ Some protocols consider congestion condition,^6,11 and others also take interference and channel utilization into consideration.¹² In terms of the different features of mesh routers and clients, some protocols are node-aware routing protocols,¹³ while some others just consider all the nodes in the network together. Although there are some routing protocols designed for hybrid WMN, the ones taking mobility and energy constraints of mesh clients into account are quite few.

Routing metric is the core measuring standard in routing protocol to choose the best route.^14,15 It can contain a series of measurements by mathematical way. The measurements, which can be obtained from media access control (MAC) layer and physical layer, are often used in the routing protocol in network layer. This article concentrates on the routing metric improvements based on the cross-layer aware (CLA) routing protocol.¹⁶ A regional energy- and mobility-aware (REMA) routing protocol is proposed. The main contributions of REMA are as follows:

The regional residual energy of neighbor mesh clients is taken into account in the routing metric. At the same time, the degree of dispersion of the remaining energy is computed. A region with more remaining energy and less degree of dispersion indicates that less packets are transmitted within this area. The node whose neighbor area has sufficient and balanced energy can be selected for network services. Energy can be balanced and the network lifetime can be prolonged.

The mobility of clients in a whole path is considered to enhance the stability and reduce network overhead. Besides the speed of a single mesh client, the mobility of clients in the whole path is also taken into account. The route with less average speed, less speed dispersion, and more stability will be chosen. In this case, frequent link breakup can be avoided and the overhead brought by discovering new route frequently can be reduced.

Related work

The number of routing protocols designed for hybrid WMN currently is not large. Ad hoc on-demand distance vector (AODV)^17,18 is the most popular routing protocol that is used in hybrid WMN to adapt to dynamic changes. However, AODV only uses hop count as the routing metric, neglecting interference, load, mobility, and energy constraints, which overlooks typical features of hybrid WMN. High-performance AODV (HP-AODV)¹⁹ and multi-link AODV (ML-AODV)¹² use multi-link and recommended channel approach to decrease interference. SafeMesh¹³ considers the queue length at different interfaces and congestion to improve the performance of hybrid WMN. Congestion-adaptive AODV (CA-AODV)⁶ based on AODV makes use of the routing metric Channel Diverse Congestion Aware (CDCA) to build a less load and interference route. CDCA gets queue length in MAC layer through the cross-layer method to measure the load. Intra-flow interference is also measured by channel diversity to improve the network performance. Mobility and energy constraints of mesh clients are neglected in CDCA, which bring less accuracy to evaluate the network condition. Also, on the basis of AODV, Adaptive Load-Aware Routing Metric (ALARM)¹⁶ describes load and interference by queue length from MAC layer and the current data transmission rate. Cross layer Secure and Resource-aware On demand Routing (CSROR)²⁰ is designed to ensure routing security. Mobility is not considered in ALARM or CSROR either. Dynamic Weighted Cumulative Expected Transmission Time (D-WCETT)²¹ modifies Weighted Cumulative Expected Transmission Time (WCETT)²² to change the weight value of $β$ . D-WCETT also obtains the degree of congestion by cross-layer approach, which can adapt to the change of topology. Modifying D-WCETT, Energy-Load Aware Routing Metric (ELARM)²³ replaces the sum of Expected Transmission Time (ETT) with the remaining energy obtained from the application layer. The selected path can be adaptive to the change of energy and load. Nevertheless, in D-WCETT and ELARM, large overhead can be caused by the proactive probe method of Expected Transmission Time (ETT), decreasing network performance.

CLA based on AODV applies hop count, channel utilization, speed, and residual energy to the routing metric. Network interference, load, mobility, and energy are considered synthetically in CLA by cross-layer method. The definition of the routing metric in CLA, that is $M_{CLA} (p)$ ) is

$\begin{matrix} M_{CLA} (p) = \sum_{link i \in p}^{n} CB T_{i} + max_{node j \in p, 1 \leq j \leq m} \frac{V_{j}}{V_{\max}} \\ + max_{node j \in p, 1 \leq j \leq m} E_{j} + \frac{Hop_count}{Hop_coun t_{max}} \end{matrix}$ (1)

where n and m are the number of links and nodes in path p, respectively. $V_{j}$ is the maximum speed in path p. $V_{\max}$ denotes the maximum speed in the whole network. With the increase in the ratio of $V_{j}$ to $V_{\max}$ , the probability of link breakup also arises. $E_{j}$ is the weight value of residual energy. $Hop_count$ denotes the path length, and $Hop_coun t_{\max}$ is the maximum count which can be forwarded in the whole network. The path length influences route stability. CBT is the time period of channel being busy and is proved as an accurate approach to measure the use efficiency of channel. Logical interference and load can also be measured accurately.²⁴CBT can be denoted as

$CBT = \frac{T_{total} - T_{idle}}{T_{total}}$ (2)

It can be seen that the metric of CLA considers channel condition, mobility, energy, and path length. The features of mesh routers and clients are also considered in CLA. Inter-flow interference, traffic load, stability, and energy validity can be measured effectively. However, CLA also have some disadvantages as follows:

The most remaining energy weight in the path is used into the routing metric, which neglects energy condition of neighbor nodes. The selected path can cause centralized consumption of energy in an area, and besides the importance of energy balance, network lifetime is also neglected.

The routing metric only considers the highest speed in the path and cannot take the influence of other nodes in the same path into overall consideration.

It is a source routing protocol, which is non-order-preserving and can cause high cost to store the path information in hybrid WMN.

REMA routing protocol

To solve the disadvantages of CLA, REMA measures the regional energy consumption and the mobility in the whole path respectively.

Regional energy-aware model

As shown in Figure 1, when source node s wishes to connect destination node d, node b and node j are packet forwarders. Nodes e, f, and p are neighbor mesh clients of node j.

Figure 1.

Schematic representation of regional energy.

The regional energy of node j is the set of energy-consumed condition of its neighbor mesh clients, which reflects the concentration degree of energy consumed. A computing method of average degree of concentration (AVDC) is used to describe this energy-consumed condition. AVDC is the sum of average energy consumption and the standard deviation of neighbor mesh clients. Its definition expression is

$\begin{matrix} AVDC (j) = \frac{1}{N} \times \sum_{k \in S (j)} E (k) \\ + {(\frac{1}{N} \times \sum_{k \in S (j)} {(E (k) - \frac{1}{N} \times \sum_{k \in S (j)} E (k))}^{2})}^{\frac{1}{2}} \end{matrix}$ (3)

where N is the number of neighbor mesh clients of node j, and S(j) is the set of neighbor mesh clients. E(k) is the intensity of energy consumption of mesh client k, which can be described by the ratio of consumed energy to initial energy of mesh client k. High value of AVDC demonstrates high degree of concentration in energy consumption, as well as traffic load in this area. Thus, data packets should not be forwarded by nodes with high regional energy. In this way, energy and load can be balanced by choosing a node with less AVDC.

Routing metric of REMA

The routing metric of REMA reserves the measurement of interference and load (i.e. CBT) in CLA, improving the measurement of energy and mobility. AVDC is used in the regional energy-aware part of the routing metric (i.e. EN(j)), and EN(j) can be denoted as

$EN (j) = En (j) + AVDC (j)$ (4)

En(j) is the intensity of energy consumption of current node j, and the expression is

$En (j) = 1 - \frac{E_{r} (j)}{E_{initial} (j)}$ (5)

where $E_{r} (j)$ and $E_{initial} (j)$ are the remaining energy and initial energy of the current node j, respectively.

According to the regional energy-aware part of REMA, the process and the analysis of routing are shown in Figure 2. From the source node S to the destination node D, the two optional routes are S-A-B-C-D and S-L-M-N-D. The sum of EN in the first path is 5, which is lower than that in the second path with the value of 7.8. EN considers both the residual energy and the regional energy condition of each node. The lower sum of EN illustrates the higher residual energy and better network condition. Therefore, the first route S-A-B-C-D is selected.

Figure 2.

Process of routing according to the regional energy-aware part of REMA.

For the mobility part, to measure client speed more accurately and attain the order preservation in the routing metric, a similar computing method of average degree of dispersion (AVDD) is adopted. AVDD is the sum of average speed and standard deviation of mesh clients in the whole path and is defined as

$\begin{matrix} AVDD (j) = \frac{1}{N} \times \sum_{k \in p} \frac{V (k)}{V_{\max} (k)} \\ + {(\frac{1}{N} \times \sum_{k \in p} {(\frac{V (k)}{V_{\max} (k)} - \frac{1}{N} \times \sum_{k \in p} \frac{V (k)}{V_{\max} (k)})}^{2})}^{\frac{1}{2}} \end{matrix}$ (6)

where N is the count of mesh clients in the path p containing current node j. V(k) is the speed of mesh client k in the path p, and $V_{\max} (k)$ is the maximum speed of mesh client k.

The mobility part of the routing metric (i.e. SD(j)) is expressed as

$SD (j) = Sd (j) + AVDD (j)$ (7)

where Sd(j) is the intensity of mobility of current node j, and it can be expressed as

$Sd (j) = \frac{V (j)}{V_{\max} (j)}$ (8)

where V(j) is the speed of node j, and $V_{\max} (j)$ is the maximum speed in the whole network.

According to the mobility part of REMA, the process and the analysis of routing are shown in Figure 3. From the source node S to the destination node D, there are also two routes which are S-A-B-C-D and S-L-M-N-D. The average value and standard deviation of v in the first path are lower than those in the second path, which means the stability of the first path is much better. Therefore, the first path S-A-B-C-D is selected.

Figure 3.

Process of routing according to the mobility part of REMA.

The routing metric of REMA in path p (i.e. MREMA(p)) is

$\begin{matrix} MREMA (p) = \sum_{link i \in p}^{m} CBT (i) \\ + \sum_{node j \in p}^{n} EN (j) + \sum_{node j \in p}^{n} SD (j) \end{matrix}$ (9)

The variables which are used in REMA routing protocol are listed in Table 1.

Table 1.

Variable definition in REMA routing protocol.

Variable	Definition
N	Count of mesh clients in the path p
S(j)	Neighbor mesh clients of node j
E(k)	Energy consumption intensity of client k
AVDC(j)	Regional energy-consumed condition of node j
En(j)	Energy consumption intensity of node j
$E_{r} (j)$	Remaining energy of node j
$E_{initial} (j)$	Initial energy of node j
EN(j)	Regional energy-aware part of routing metric
V(j)	Speed of mesh client j
$V_{\max} (j)$	Maximum speed of mesh client j
AVDD(j)	Speed dispersion of a path including node j
Sd(j)	Mobility intensity of node j
SD(j)	Mobility part of the routing metric
CBT(i)	Channel information of link i
MREMA(p)	Routing metric of REMA in the path p

From mentioned above, we can see that REMA routing protocol has several advantages:

Regional energy is considered in the route discovery process, which balances energy and load. The network’s lifetime can also be prolonged.

The measurement of mobility is improved, and the speed of mesh clients in the whole path is taken into account. The selected route is more stable.

The routing metric of REMA routing protocol has order preservation, decreasing the network cost.

After a node receives a route request (RREQ) message, the processing mechanism is shown in Figure 4.

Figure 4.

RREQ processing mechanism.

Computation complexity of REMA

Assuming the number of nodes in a whole path is n. The computation complexity of REMA depends on formula (9) which is composed of three parts. For the first part of the formula, every link in a path should compute a value of CBT, and the computing time of CBT in a whole path is n–1. For the second part, the regional energy-consumed condition is computed as equation (4) at every node in a path. The computation complexity of this process depends on formula (3). A path containing n nodes will compute n times. For the third part of formula (9), the computation complexity is similar to the second part. The speed dispersion condition of a path is computed. Each node in a path should run formula (7) for one time, and the computation complexity of formula (7) depends on formula (6). A path containing n nodes will compute n times. In conclusion, the computation complexity of the routing metric of REMA is according to the most computation complexity of the three parts. Therefore, the computation complexity of REMA is O(n). In the worst case, if the number of nodes in the hybrid WMN is assumed to be m, the most count of nodes in a path is m. In this case, the computation complexity of REMA is O(m).

Performance evaluation and simulated analysis

NS2 network simulator²⁵ is used to compare the performance of REMA, CLA, ALARM, and AODV.

Simulation parameters

At first, simulation models in NS2 are extended to multiradio–multichannel environment based on Roman scheme.²⁶ The backbone of hybrid WMN is deployed into a 5 × 5 grid or a random topology. The size of the simulation area is 1000 m × 1000 m. Detailed simulation parameters are shown in Table 2. The number of data flows and maximum speed of clients are changed to evaluate network performance.

Table 2.

Detailed simulation parameters.

Simulation parameters	Values
PHY/MAC technology	IEEE 802.11b
Nominal bandwidth	2 Mbps
Data flows	Constant bit rate
Packet size	512 bytes
Packet rate	80 kbps
Number of mesh routers	25
Number of mesh clients	50
Number of interfaces in each mesh router	3
Number of channels in each mesh router	3 channels
Number of interfaces in each mesh client	1
Number of channels in each mesh client	1 channel
Initial energy of each router (J)	10,000
Initial energy of each client (J)	100
Transmission range	250 m
Interference range	550 m
Propagation model	Two-ray ground
Mobility model	Random way point
Antenna	Omnidirectional
Simulation time	900 s

Simulation parameters

Average network throughput: the bit number successfully received by destination nodes per second. The unit is kilobit per second (kbps).

Average packet loss rate: the ratio of packet number unsuccessfully received by the destination to the packet count sent by the source.

Average end-to-end delay: the ratio of the total time delay to the packet number successfully received by the destination. The unit is second per packet (s/p).

Average routing overhead: the ratio of control packet number to the count of data packets successfully received by the destination.

Energy consumption of every data packet: the ratio of the total consumed energy to the number of packets received by the destination. The unit is Joule per packet (J/p).

No-balance degree of remaining energy in mesh clients: the standard deviation of remaining energy of all mesh clients in the whole network when the first mesh client becomes invalid due to the exhausted energy.^27,28

Network lifetime: the run time until the first client runs out of energy. The unit is seconds.

Simulated analysis

Performance comparison under different number of flows in hybrid WMN with grid backbone topology

The backbone composed of mesh routers in hybrid WMN is deployed into a 5 × 5 grid, and mesh clients are deployed randomly. The maximum speed of mesh clients is fixed to $10 m / s$ , and the number of data flows increases. The performance comparisons between REMA, CLA, ALARM, and AODV in terms of the seven performance metrics are shown in Figures 5 –11.

Figure 5.

Average network throughput versus the number of flows in hybrid WMN with grid backbone.

Figure 6.

Average packet loss rate versus the number of flows in hybrid WMN with grid backbone.

Figure 7.

Average end-to-end delay versus the number of flows in hybrid WMN with grid backbone.

Figure 8.

Average routing overhead versus the number of flows in hybrid WMN with grid backbone.

Figure 9.

Energy consumption of every data packet versus the number of flows in hybrid WMN with grid backbone.

Figure 10.

No-balance degree of remaining energy in mesh clients versus the number of flows in hybrid WMN with grid backbone.

Figure 11.

Network lifetime versus the number of flows in hybrid WMN with grid backbone.

Figure 5 shows that the average network throughput of REMA is higher than other routing protocols. Figures 6 and 7 illustrate that the average packet loss rate and average end-to-end delay of REMA are always lower than CLA, ALARM, and AODV. As more energy consumption indicates more load, REMA also measures the regional load along with regional energy. The selected route can effectively avoid an area with heavy load, so the average packet loss rate and end-to-end delay are declined. Besides, the mobility of mesh clients is also considered in REMA. The stability of the path is improved and frequent breakup caused by fast speed is averted. The packet loss rate and delay are further reduced, and the throughput is improved.

The advantage of REMA in low overhead and high energy efficiency can be seen in Figures 8 and 9. Besides the measurement of load, energy, and mobility in the routing metric of REMA, REMA also has order preservation. However, CLA is non-order-preserving, which brings higher overhead than REMA. The overhead in ALARM is also higher resulting from multiple copies of the route request packet are forwarded to build multiple links. In terms of energy efficiency, REMA can measure intensity of energy consumption in an area more accurately, and high energy efficiency can be obtained due to the energy balance. In addition, low network overhead also contributes to the energy saving in REMA.

Figure 10 shows that the no-balance degree of remaining energy of REMA is lower than CLA, ALARM, and AODV. Because of the multi-link metric in ALARM, the load can be balanced in some extent and the performance of ALARM is better sometimes. The average performance of REMA is still better than that of CLA and ALARM. Figure 11 illustrates that the lifetime of REMA is longer than CLA, ALARM and AODV. REMA describes the regional energy, which can achieve energy balance in the whole network. The energy balance can avoid large differences among energy consumption of mesh clients and then the network lifetime can be prolonged.

Performance comparison under different maximum speed of mesh clients in hybrid WMN with grid backbone topology

The hybrid WMN is also with a 5 × 5 grid backbone topology, and mesh clients are deployed randomly. The number of data flows is fixed as 11, and the maximum speed of mesh clients gradually increases. The performance comparisons between REMA, CLA, ALARM, and AODV in terms of the seven performance metrics are shown in Figures 12 –18.

Figure 12.

Average network throughput versus the maximum speed of mesh clients in hybrid WMN with grid backbone.

Figure 13.

Average packet loss rate versus the maximum speed of mesh clients in hybrid WMN with grid backbone.

Figure 14.

Average end-to-end delay versus the maximum speed of mesh clients in hybrid WMN with grid backbone.

Figure 15.

Average routing overhead versus the maximum speed of mesh clients in hybrid WMN with grid backbone.

Figure 16.

Energy consumption of every data packet versus the maximum speed of mesh clients in hybrid WMN with grid backbone.

Figure 17.

No-balance degree of remaining energy in mesh clients versus the maximum speed of mesh clients in hybrid WMN with grid backbone.

Figure 18.

Network lifetime versus the maximum speed of mesh clients in hybrid WMN with grid backbone.

Figures 12 –18 illustrate that REMA can maintain better performance with increasing maximum speed of mesh clients. REMA can select a path with less average mobility, which enhances the stability of the route. Although CLA uses the maximum speed in a path into the routing metric and aims to decrease breaking probability, CLA neglects the mobility condition of mesh clients in the whole path. ALARM does not consider the mobility features of mesh clients. AODV only considers the hop count during the process of selecting routes, which neglects the features of mesh routers and mesh clients in hybrid WMN. Therefore, REMA can choose path with higher stability. Meanwhile, regional energy-consumed condition is considered in REMA to balance energy and load. This is the reason why REMA has more advantages in hybrid WMN.

Performance comparison under different number of flows in hybrid WMN with random backbone topology

To further verify the effectiveness of REMA, the performance of REMA is also analyzed in hybrid WMN with random backbone topology. In this part, mesh clients and mesh routers are both deployed randomly. The maximum speed of mesh clients is fixed to 10 m/s, and the number of data flows is changed. The performance comparisons between REMA, CLA, ALARM, and AODV in terms of the six performance metrics are shown in Figures 19 –24.

Figure 19.

Average packet loss rate versus the number of flows in hybrid WMN with random backbone.

Figure 20.

Average end-to-end delay versus the number of flows in hybrid WMN with random backbone.

Figure 21.

Average routing overhead versus the number of flows in hybrid WMN with random backbone.

Figure 22.

Energy consumption of every data packet versus the number of flows in hybrid WMN with random backbone.

Figure 23.

No-balance degree of remaining energy in mesh clients versus the number of flows in hybrid WMN with random backbone.

Figure 24.

Network lifetime versus the number of flows in hybrid WMN with random backbone.

It can be seen from Figures 19 –24 that the performance of REMA remains better than other routing protocols in hybrid WMN with random backbone. Based on CLA, REMA measures regional energy to balance energy and load. The packet loss rate and delay can be declined. Besides, REMA measures the mobility of mesh clients more accurately, which makes the chosen route more stable. The routing metric of REMA has order preservation to reduce the occupation of network resource. Network overhead and energy consumption are reduced. With the comprehensive consideration of load, energy constraints, and mobility, REMA can balance energy, prolong network lifetime, and improve the whole network performance.

Conclusion

To adapt the features of energy constraints and mobility in hybrid WMN, a REMA routing protocol is proposed in this article. Based on CLA, REMA measures energy and mobility more accurately. In the regional energy part of REMA, energy-consumed intensity and the discretization degree of neighbors are described. The energy and load can be balanced better for the whole network, and the network lifetime is prolonged. At the same time, the mobility of mesh clients in the whole path is measured to strengthen stability. Simulation results show that REMA can achieve better network performance and obtain the goal of prolonging network lifetime, balancing energy and load.

The single-radio mesh clients are taken into account currently. We plan to extend the mesh clients to be multiple-radio, improving their data forwarding ability. Then, the performance of current routing protocols designed for hybrid WMN can be evaluated in the future.

Footnotes

Academic Editor: Salvatore Serrano

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research,authorship,and/or publication of this article: This work was supported by the National Natural Science Foundation of China (no. 61373124).

References

Akyildiz

Wang

Wireless mesh networks: a survey. Comput Netw 2005; 47(4): 445–487.

Bargaoui

Mbarek

Togni

. Hybrid QoS based routing protocol for inter and intra wireless mesh infrastructure communications. Wirel Netw 2015; 22: 2111–2130.

Avula

Lee

Yoo

SM.

Security framework for hybrid wireless mesh protocol in wireless mesh networks. TIIS 2014; 8(6): 1982–2004.

Peng

. A new dynamic hierarchical reputation evaluation scheme for hybrid wireless mesh networks. Comput Electr Eng 2014; 40(2): 663–672.

Anita

Suresh

RM.

Improving QoS routing in hybrid wireless mesh networks, using cross-layer interaction and MAC scheduling. Cybern Inform Tech 2015; 15(3): 52–67.

Pirzada

Wishart

Portmann

Congestion aware routing in hybrid wireless mesh networks. In: Proceedings of the 15th IEEE international conference on networks (ICON), Adelaide, SA, Australia, 19–21 November 2007, pp.513–518. New York: IEEE.

Nguyen

Min

Matson

. Using mobile robots to establish mobile wireless mesh networks and increase network throughput. Int J Distrib Sens N 2012; 8: 614532.

Taneja

Kush

A survey of routing protocols in mobile ad hoc networks. Int J Innov Manage Tech 2010; 1(3): 279–285.

Xie

Wang

A survey of mobility management in hybrid wireless mesh networks. IEEE Network 2008; 22(6): 34–40.

10.

Ding

Pongaliur

Xiao

Channel allocation and routing in hybrid multichannel multiradio wireless mesh networks. IEEE T Mobile Comput 2013; 12(2): 206–218.

11.

Pirzada

Wishart

Portmann

. ALARM: an adaptive load-aware routing metric for hybrid wireless mesh networks. In: Proceedings of the thirty-second Australasian conference on computer science (vol. 91), Wellington, New Zealand, 1 January 2009, pp.37–46. Darlinghurst, NSW, Australia: Australian Computer Society, Inc.

12.

Pirzada

Wishart

Portmann

Multi-linked AODV routing protocol for wireless mesh networks. In: Proceedings of the global telecommunications conference, Washington, DC, 26–30 November 2007, pp.4925–4930. New York: IEEE.

13.

Pirzada

Portmann

Wishart

. SafeMesh: a wireless mesh network routing protocol for incident area communications. Pervasive Mobile Comput 2009; 5(2): 201–221.

14.

Yang

Wang

Kravets

Designing routing metrics for mesh networks. In: Proceedings of the IEEE workshop on wireless mesh networks (WiMesh), Santa Clara, CA, 26 September 2005, pp.1–9.

15.

Campista

MEM

Esposito

Moraes

. Routing metrics and protocols for wireless mesh networks. IEEE Network 2008; 22(1): 6–12.

16.

Shi

Liu

Wang

. Cross-layer aware routing protocol for hybrid wireless mesh networks. J Commun 2015; 10(7): 480–489.

17.

Pirzada

Portmann

Indulska

Performance analysis of multi-radio AODV in hybrid wireless mesh networks. Comput Commun 2008; 31(5): 885–895.

18.

Perkins

Belding-Royer

Das

Ad-hoc on-demand distance vector (AODV) routing. RFC 3561, 2003, https://tools.ietf.org/html/rfc3561

19.

Pirzada

Portmann

. High performance AODV routing protocol for hybrid wireless mesh networks. In: Proceedings of the 4th international conference on mobile and ubiquitous systems: networking & services, 6–10 August 2007. Washington, DC: IEEE.

20.

Khan

Loo

Cross layer secure and resource-aware on-demand routing protocol for hybrid wireless mesh networks. Wireless Pers Commun 2012; 62(1): 201–214.

21.

Ali

Kiani

Pirzada

AA.

Load dependent dynamic path selection in multi-radio hybrid wireless mesh networks. In: Proceedings of the IEEE wireless communications and networking conference (WCNC), Istanbul, 6–9 April 2014, pp.2020–2025. New York: IEEE.

22.

Draves

Padhye

Zill

Routing in multi-radio, multi-hop wireless mesh networks. In: Proceedings of the 10th annual international conference on mobile computing and networking, Philadelphia, PA, 26 September–1 October 2004, pp.114–128. New York: ACM.

23.

Kiani

Ali

Rashid

Energy-load aware routing metric for hybrid wireless mesh networks. In: Proceedings of the IEEE vehicular technology conference (VTC Spring), Glasgow, 11–14 May 2015.

24.

Borges

VCM

Curado

Monteiro

Cross-layer routing metrics for mesh networks: current status and research directions. Comput Commun 2011; 34(6): 681–703.

25.

Issariyakul

Hossain

Introduction to network simulator NS2. Berlin: Springer, 2011.

26.

Calvo

Campo

JP.

Adding multiple interface support in NS2, pp.21–35, 2007, http://personales.unican.es/aguerocr/files/ucMultiIfacesSupport.pdf

27.

Jiang

Sun

Energy optimized routing algorithm for hybrid wireless mesh networks in coal mine. Int J Distrib Sens N, 2015; 11: 237697.

28.

Peng

Guo

. A new energy-efficient on-demand routing protocol for green wireless mesh networks. Adv Mat Res 2013; 756: 2275–2280.

Regional energy- and mobility-aware routing protocol for hybrid wireless mesh network

Abstract

Keywords

Introduction

Related work

REMA routing protocol

Regional energy-aware model

Routing metric of REMA

Computation complexity of REMA

Performance evaluation and simulated analysis

Simulation parameters

Simulation parameters

Simulated analysis

Performance comparison under different number of flows in hybrid WMN with grid backbone topology

Performance comparison under different maximum speed of mesh clients in hybrid WMN with grid backbone topology

Performance comparison under different number of flows in hybrid WMN with random backbone topology

Conclusion

Footnotes

Declaration of conflicting interests

Funding

References