The wireless mesh network (WMN) has proven to be a great choice for network communication technology. WMNs are composed of access points (APs) that are installed and communicate with each other through multihop wireless networks. One or more of these APs acts as a gateway (GW) to the internet. Hosts of WMNs are stationary or mobile. According to the structure of WMNs, some network features may be affected, such as the overall performance, channel interference, and AP connectivity. In this paper, we propose a new adaptive channel allocation algorithm for a multiradio multichannel wireless mesh network. The algorithm is aimed to minimize the number of channel reassignments while maximizing the performance under practical constraints. The algorithm defines a decision function for the channel reassignments. The decision function aims to minimize the traffic around the GW. Whenever the traffic changes in the wireless mesh network, the decision function decides which channel radio reassignment should be done. We demonstrated the effectiveness of our algorithm through extensive simulations using Network Simulator 2 (NS2).
Over the past decade, the fifth generation (5G) and Internet of Things (IoT) network systems and applications have been developed extensively to deliver a reliable, low latency, real time connectivity network [
At least one AP in a WMN operates as a gateway (GW) to the internet. A host can reach the internet through one of the GWs after connecting to it in multihop wireless communications between intermediate APs. Subsequently, wireless links around GWs are normally quite congested, and this may lead to a bottleneck in the entire communications of WMN.
In WMN, each AP performs two distinct roles: one of these roles is connecting hosts to the network, and the other is conveying communications between APs. Therefore, for large scale WMN, it is essential to maintain the performance and decrease traffic toward GWs as the related hosts are increased [
WMN has many issues to be resolved before being deployed due to the basic network essential requirements for everyday use [
Another important issue that faces WMNs is the channel allocation problem. The efficient utilization of constrained channels to manage the increase in traffic capacity requires the best possible channel assignment. This concept is a crucial issue to achieve a stable WMN. The two methodologies for allocating channels to APs are fixed channel allocation (FCA) and dynamic channel allocation (DCA) [
Our objective in this paper is to propose a new adaptive channel allocation algorithm for a multiradio, multichannel wireless mesh network with a practical constraint. The practical constraint is on the number of network interface cards (NICs) that can be installed per AP. In this algorithm, we characterize a decision function for the channel reallocation. Our algorithm is separated into two phases: a static assignment phase and an adaptive assignment phase. In the static (initial) phase, an appropriate, fixed number of network interface cards (NICs) with appropriate channels are assigned to APs, utilizing the connection with the most extreme number of hosts related to each AP, such that the performance is maximized while the maximum number of hosts per AP is minimized. In the adaptive phase, the decision function chooses whether the channel reallocation must be performed or not, guided by the traffic balance between the links to the GW. The decision function is also intended to minimize the number of channel reassignments to reduce the turn off time of the network tasks during channel reallocation methods. The effectiveness of our proposed algorithm was verified through Network Simulator 2 (NS2) simulations. We used our modified 802.11n module on the NS2 version 2.34 presented in [
This paper is organized as follows, In
One of the greatest challenges for deploying a sustainable wireless mesh network is the channel allocation problem. The overall performance of the network can be increased by allocating suitable channels to minimize the network interference. One of the greatest challenges for deploying a sustainable wireless mesh network is the channel allocation problem. The overall performance of the network can be increased by allocating suitable channels to minimize network interference. Wang et al. [
In [
The authors in [
In [
The authors in [
The authors in [
The objectives of this problem aim to maximize the network performance and to minimize the channel reallocation. In the following section, we introduce the assumptions and mathematical model of the problem under study.
A wireless link may interfere with other wireless links due to the broadcast nature of the links, and this may happen if the links are within transmission range of each other. Interfered links can not transfer data at the same instant of time if they use the same channel. The physical interference model [
In this section, we formulate the mathematical model for the network. Suppose that the WMN is described as graph topology
Let
To utilize the maximum throughput through the network, we calculate the traffic between APs through links, and suppose that
In the proposed model, the number of distributed NICs per each AP is calculated through our procedure. Let
Suppose that
For the standard of the IEEE802.11ax protocol with the orthogonal frequencydivision multiplexing (OFDM) [
In this work, we suppose that each AP is associated with the most extreme number of hosts as we designed the network according to the greatest burden for each AP. To compute the total number of NICs appointed to each AP in the whole network, certain critical physical constraints should be applied to the algorithm, and these constraints are as follows:
The total number of NICs over the network should not exceed
The output number of assigned NICs for
Each link
The maximum number of assigned channels is
The chosen channel for the link
The number of channels assigned to the links adjacent to
The number of NICs assigned to
The channel between the two APs
To achieve a higher throughput and meeting minimum cost, our proposal intends to minimize the following functions:
Throughout this section, we introduce a method to solve the channel allocation problem dynamically but with adaptive control. The adaptive channel allocation algorithm is separated into two parts, the initial channel allocation (installation phase), and the adaptive channel allocation (running phase).
The initial channel allocation draws out the number of NICs installed per AP in the network and determines the channel allocation as a start up phase. According to this part, the traffic of each AP is assumed to be the maximum associated load.
For the adaptive channel allocation phase, we introduce a decision function to determine whether the network should reallocate the channel or continue with the last allocation.
Through this section, an initial allocation of the channels is built, considering the case where the network is fully loaded. This phase consists of two stages, the NIC deploying stage and the fixed channel allocation stage to solve the channel allocation problem in the design mode for multiradio multichannel WMN. This configuration is considered as FCA, in which the output is a static channel allocation. In wireless mesh networks, the APs topology is fixed; however, APs must collect information regarding the hosts to deploy the initial stage, such information collected by the
Through this stage, one NIC is deployed to an AP subjected to the maximum traffic load per previously assigned (
Compute the load per AP from link traffic
Allocate one NIC to every AP:
Assign the traffic for each NIC by:
Set the number the allocated NICs: AN = N.
Add one NIC to an AP (suppose
Stop the phase if
Recalculate the traffic:
Return to step (i).
The output of the fixed channel allocation phase is a set of channels assigned to the deployed NICs in
Compute the collision for the traffic:
Arrange the links according to the collision in a descending order.
Allocate
Allocate the links’ channel according to the following filtration steps:
Allocate
If
For each link
if
In this phase, we introduce a new decision function to decide whether to change the channel allocation dynamically or to use the same allocated channel. The concept behind the decision function originates from the limited number of links adjacent to the GW, in which the maximum traffic must pass through those links to explore the internet. The congestion around the GW’s links may cause bottlenecks; therefore, the traffic over these links should be distributed evenly to avoid such a situation. The channel reallocation should occur exactly when the uniformity of the traffic is not fulfilled by the last channel allocation.
The decision function is presented by the following steps.
We begin by calculating the traffic of GW’s links, where these links are assigned the same channel
We compute the decision factor:
where the parameter
The decision function (DF) is defined as follows:
According to the result of the decision function, the channel reallocation is utilized by using the same steps in
Throughout this section, we examine our algorithm using our modified NS2 presented in [
Network Simulator 2 (NS2), is an opensource network simulator. We implement modifications to the physical and MAC layers for NS2 to achieve a simulator for IEEE 802.11n through the framework NS2.34 Version [
The physical layer of NS2.34 802.11 is an easy model to understand in which the improved version is embedded in [
In the MAC layer, we embed the IEEE 802.11n MAC protocol files as a new MAC protocol type for the NS2.34. We implement only the aggregation MAC protocol data unit aggregation (AMPDUA) into the MAC layer. Normally, AMPDUA is less efficient than the aggregation MAC Service Data Unit (AMSDU) aggregation. However, AMPDUA can be more efficient in environments with high error rates due to the use of the block acknowledgment mechanism. The acknowledgment of every aggregated data frame is done separately, and to be retransmitted if suffering from error. In the interface queue type, we implement the new aggregation module.
The simulation is done by assuming that each host has full constant bit rate (CBR) traffic coordinating to the related AP. The length of each data packet handed to the MAC layer is 1000 bytes. We suppose that the frame size of the MAC layer is a maximum of 8 Kbytes as an aggregated payload. The remainder of the NS2 simulation assumptions are shown in
Three distinctive schemes are simulated through this work. The first one is the simulation of static channel allocation, in which we assign static channels to the APs after associating the most extreme number of hosts related to each AP. In other words, the channels are fixed regarding any traffic changes.
The second schema is the simulation of a fully dynamic change channel assignment; every time the traffic is changed, the channel reassignment is done. The third schema is the proposed adaptive channel allocation one, in which the decision function decides whether to change the channel assignment or maintain the channel assignments as is.
Through the simulations, we assume that the maximum number of hosts expected to attach to every AP is provided as an input. The hosts are distributed among 24 different times for every AP with random distribution from one of the seven host association patterns, which are illustrated in
The simulations are conducted for two network topologies, illustrated in
The first simulation takes place for the network topology 1,
For example, concerning the first value of
As a discussion of the illustrated results, we can choose the throughput despite the stopping time caused by the dynamic reassignment. However, this depends on the application that runs over the connected channel, applications, such as voice over internet protocol (VOIP) and video or voice streaming cannot compromise the stopping slots with the average throughput, which demand a need for balancing between the average throughput and the reassignment times.
We can see that the
The second simulation was performed for network topology 2, 5 × 5 (
This case is performed using at most two NICs installed per each AP.
In this paper, we presented the formulation of the adaptive allocation problem for multiradio, multichannel wireless mesh networks (WMN) with a practical constraint on the number of NICs per AP. The adaptive allocation algorithm is made of the initial phase and the dynamic phase, the dynamic phase introduces a newly defined decision function. The viability of our methodology is checked through our modified version of the NS2 network simulation. The notable enhancement of the performance is seen by picking a suitable
Conceptualization, W.H. and T.F.; methodology, W.H.; validation, W.H. and T.F.; formal analysis, W.H. and T.F.; investigation, W.H. and T.F.; resources, T.F.; data curation, W.H.; writing—original draft preparation, W.H.; writing—review and editing, T.F.; visualization, W.H. and T.F. All authors have read and agreed to the published version of the manuscript.
This research received no external funding.
The authors declare no conflict of interest.
The following abbreviations are used in this manuscript:
IEEE 802.11n implementation in Network Simulator (NS)2.34.
Patterns of numbers of the associate hosts, supposing that the maximum number of associated hosts is n.
Simulation network topologies.
The network topology 1 results, the number of channel reassignments, and the overall throughput with the maximum two NICs, (
The network topology 1 results, the number of channel reassignments, and the overall throughput with the maximum three NICs, (
The network topology 2 results, the number of channel reassignments, and the overall throughput with the maximum two NICs, (
The network topology 2 results, the number of channel reassignments, and the overall throughput with the maximum three NICs, (
Formulation notations.
Symbol  Definition 


link between the access point (AP) 
NOC  nonoverlapped channels 

interference matrix between links 

expected throughput between 

the matrix of interfered channels 

the maximum number of network interface cards (NICs) that can be allowed to 

the output assigned number of NICs to 

the assigned channel to 

maximum number of hosts per one NIC 

total interference of gateway (GW) adjacent links 

number of channel reallocations 
NS2 Parameters.
Contention window_Min  15 
Contention window_Max  1023 
Time Slot  9 
Short Interframe Space  16 
DCF Interframe Space  34 
Preamble Length  16 
Physical Layer Convergence Protocol Header  48 bits 
Physical Layer Convergence Protocol Rate  6 Mbps 
Basic Data Rate  54 Mbps 
Data Rate  300 Mbps 