MAC layer has a responsibility to access the node coordinates through shared channel which in terms it reduce the conflicts. In this context, an efficient channel allocation scheme via multi-hop network designed by several researchers and its intention behind the study are discussed below,
Resource allocation scheme is presented by K. Shashi Raj et al. [12] who aims to meet better QoS using Interference Resilient Stochastic Prediction technique. This method is intend to reduce the interference made by secondary user whereas provide greater amount of resource utilization. With this allocating technique, the noise variance and interference is being reduced to access the resource with obtaining high QoS.
In dynamic channel allocation, topology infrastructure of MANET is unpredicted in advance due to change of node mobility and also congestion between the nodes is much complicated to predict. Routing based on centralized technique is not to be applied directly in joint channel allocation. Hon Sun Chiu et al. [13] presented a joint channel assignment and routing (J-CAR) scheme utilized mainly to reduce the interference and collision within the channel. This concept jointly interfaces two procedures of channel allocation and routing to reach bottleneck link. Smallest interference channel is assigned from local optimization process and the constraint least path is employed by widest path routing. This routing is adjustable to considered only the shortest path by terribly bypass the long route. As comparably demonstrate the channel performance, J-CAR achieves better optimal result of good throughput and very low packet delay. In previous literature, several optimization techniques of channel allocation schemes are employed to improve the network performance of MANET, cellular network etc., [14 –16]. The recommended channel allocation techniques optimized using graph-theoretic, machine learning have been suffered from hidden terminal problem. To solve this obstacle, a cluster based channel allocation techniques such as Ant colony optimization (ACO), particle swarm optimization (PSO),imperialist competitive algorithm (ICA), genetic algorithm (GA) and other several meta-heuristics are implemented to assign the channel.
Mahboobeh Parsapoor et al. [17] proposed grouping ICA (GICA)meta-heuristics technique which is capable of allocating unused channel within the cluster by effectively degrade the co-channel interferences. It is a clustered topology suggests to represents the channel as province and resource. These two terms are the primary operators that converge independently to obtain global result. To notify the capability of GICA, the investigation shows that it use less usage of channel and faster convergence than the grouping genetic algorithm (GGA). An advance imperialist competitive algorithm is proposed by Seyyed-Mohammad Javadi-Moghaddamet al. [18] who computes the heterogeneous resources at distinct time interval. This paper proposed an efficient resource allocation model by combining Tabu search with ICA. Both this searching norm scan possibly able to allocate the resource at reduced time which in terms early convergence takes place. Nevertheless, the hybrid technique suffers from network complexity problem. Sureddi Rama Koti Murali Krishna et al. [19] presented approach a hybrid approach of PSO and GA which reliably optimized to detect secure path transmission in MANET. Being hybrid can console to provide optimum result but suffer from local optima problem. To override the problem, the researchers were used several meta-heuristic approaches to solve these difficulty in network area.
System Model:
In this section, the new channel allocation scheme is proposed to avoid co-channel interference problem in MANET. Distributed network of MANET has many channels in each cell and is exchanged their message across the cell towers in an on demand basis. Nonetheless, the network is more acceptable to cause interference problem and less channel reuse efficiency. This complication is actually occurred in presence of node mobility and it is overwhelmed the network performance at greater extend. So, a generalized model of distributed channel allocation scheme is proposed by optimizing hybrid architecture in it. This tendency able to compute optimal channel allotment which in terms the technique attains high channel reuse possibility.
Hybrid optimizer of memory dragonfly algorithm (DA) and imperialist competitive algorithm (ICA) are jointly optimized in distributed channel allocation (DDCA) scheme to compute near-optimal allocated channel. This combined optimizer is named as hybrid memory dragonfly with imperialist competition (HMDIC) model. This technique guides to search the available channel in an effective manner.
Figure 1, shows the system model of proposed DDCA scheme. It consists of number of mobile nodes that can be moves randomly within the cell. Here A, B,..,I are the mobile nodes that can frequently access the network through channels. Sometime, the mobile node acts as a router to make secure channel allocation task. In figure 1, the mobile node ‘A’ dial a call to mobile node ‘D’, at that time, HMDIC technique is implemented between the nodes to identify whether there is any available channel present in the cell. Dashed line denotes the available channel that would be route through ‘B’. Next, the general description of genetic and imperialist competition algorithm are discussed as follow,
Dragonfly Algorithm:
It is a searching approach purposefully intends to analyse the behavior of dragonflies. Dragonflies are grouped together to perform two strategies namely migration and hunting. Both these strategies are comes under static and dynamic swarm behavior. Static behavior is represented by creating a minor dragonflies group that abruptly moves to hunt the prey locally. Conversely, the dynamic behavior is represented by migrated the whole dragonflies group that flies over longer distance. In search space, the optimization technique balance two phases of exploration and exploitation to reach the global optimal solution. To construct dragonfly approach, it follows four procedures to find the optimal solution i.e., separation, alignment, cohesion, attraction towards prey and distraction away from enemies. In separation, each individual move forward to avoid collision with their neighbourhood. In alignment, each individual is periodically updated the current velocity position with respect to its neighbour individual. In cohesion, all the individuals have a tendency to follow the center point that denotes the attraction and repulsion result of optimal value. If the center detection is their prey then all the individuals are attracted towards the centre or else it is their enemies then they distract away from the center. For dragonfly implementation, first of all we have to consider population size ‘n’ of dragonfly and then summarized its corresponding position. The described statement is mathematically expressed as,\({Y}_{j}={y}_{j}^{1},{y}_{j}^{s}\) ….\({y}_{j}^{n}\)where, \(j\in (1,..n)\)indicates the position of each individual with ‘n’ population size, \({y}_{j}^{s}\) specifies \({j}^{th}\) position of each individual in \({s}^{th}\)dimensional search space. For each individual, the fitness value has to be evaluated so that continuous updating of position and velocity is made possible for separation, alignment and cohesion factors [20]. The respective processing factors are mathematically derived as,
Where, \(Y and n\) are current position of each individual and population size of dragonflies respectively, \({v}_{j}\) specifies the velocity value of \({j}^{th}\) individual and \({Y}_{j}\) specifies the position value of \({j}^{th}\) individual. The mathematical equation computed for attraction (\({A}_{j}\)) and distraction (\({D}_{j}\)) logic is given by,
Where, \({Y}_{++} and {Y}_{--}\) are represented by food and enemy position at center point respectively. The aforesaid possibilities are computed by taken Euclidean distance between each individual and its direct neighbour. The distance calculation is derived below,
$${ed}_{ji}=\sqrt{\sum _{k=1}^{s}{\left({y}_{j,k}-{y}_{i,k}\right)}^{2}}$$
6
The process continues until the global position is reached. For those evaluation criteria, each time the individual updated its current velocity and position to globally improve the fitness calculation. As a result, an improved optimal result is obtained at the end session.
Imperialist Competitive Algorithm:
ICA is an evolutionary algorithm that comes under the metaheuristic optimization scheme. The concept of imperialist competitive algorithm is inspired by taken competition between two empires. As a result, a powerful strong empire is established to meet an optimal result. The technique is very competitive to capture each country and analyse its cost function, so that the empire has to be strongly build. Initially, an individual population referring in this algorithm is actually a ‘country’ and it is stated as,
$$country={country}_{1},{country}_{2},.. {country}_{n}$$
7
At first, an individual population holds the set of countries with distinct optimization parameter values. Then the power value is computed for each country and its objective function is defined as,
$${power}_{j}=\left[\frac{{norm\_cost}_{j}}{\sum _{j=1}^{n}{norm\_cost}_{j}}\right]$$
8
Where, \({norm\_cost}_{j}={cost}_{j}-minimum\left(cost\right)\) is the objective formulation for normalized cost function and \(\text{'}cost\text{'}\) term shows the cost value for entire countries individually and is represented by,
$$cost={cost}_{1}\dots {cost}_{j}\dots {cost}_{n}$$
9
As per the power calculation, the individual are categorized under two sets namely colonies and imperialist. Imperialist is actually termed as countries that possess low power colonies and enlarge its ruling coverage as higher power. So that empire has to be formed.
Final sequencing process is evolutionary operators which continuously updating the current power value of each individual changing characteristics of colonies. The process stops to find global optimal solution when there is single empire with highest power.
Assimilation, revolution, exchange and competition are the evolutionary factors explained [21] as follows,
Assimilation operator:
This operator has a tendency to change the characteristics of weakest colony of each empire and updates its corresponding cost function.
Revolution operator:
Revolution operator is used in the imperialist of weakest empire which in terms it changes its behavioral parameter. This functionality is better employed to avoid the local optimum problem of individual by just adjusting its parameters.
Exchange operator:
Exchange operator is much useful to match the power value of imperialist and their corresponding colonies of each empire. It continuously exchanges the current position of colonies with respect to their corresponding imperialist. The goal of this operator is to find low power imperialist than the respective colonies.
Competition operator:
In competition, the weakest power ensuring colony is picked up and merges it with another empire to form strong imperialist. This competitive process deployed to increase the strong imperialist cost by just eliminates the weakest one.
Distributed channel allocation designed using HMDIC:
In distributed channel allocation, the hybrid architecture of memory dragonfly and imperialist competitive algorithm are designed to find the global optimal channel as an allocated channel for the sender’s call request. Mobile ad hoc network in DDCA, shared large number of channels within the cell which predominantly independent to allocate suitable communication between two nodes. Firstly, node ‘A’ make a call request to node ‘D’, then the accessing medium that is individuals from dragonfly algorithm continuously searches the entire channel to find out the locally pbest solution of each individual node. The searching procedure is evaluated using eqn (1-5) and then continuously updated its fitness function. If the fitness value is less than pbest solution, then update the pbest solution and stored them into matrix table. Otherwise, the corresponding channel is busy so move on to the next channel to continue the same process.
While searching, DA easily stuck into premature convergence as a result, they fall into local optima problem. To escape from this problem, DA used internal memory to keep track the co-ordinates in the search space. MDA overridden the difficulties to update the local best solution of each individual and store them effectively. After completed the updating procedure, imperialist competition takes place. It performs some ruling procedure to finally detect the global solution as most powerful imperialist.
In this stage, the position of pbest solutions is assigned as colony and defined it as channel. The power of imperialist has to be computed, so that we have to create an empire. The power calculation is computed based on the cost function of each colony. So this technique performs evolutionary operation to find the powerful imperialist and it is discussed in ICA section.
In assimilation operator, the cost value is computed from each empire’s weakest colony and in revolution operator, tuning behavior can appropriately changes the imperialist’s weakest empire. It states that the condition of colony has less cost than the imperialist then change the imperialist position by that of colony. Then, compute cost value of all empires and apply imperialist competition. This competition is being established to join the weakest colony to the powerful imperialist. For that reason, the strong imperialist is formed as a global optimal solution. Now the call reached its optimal allocated channel then node ‘D’ attend the call from node ‘A’. The condition is not satisfied, then call is blocked and it again iterates the same ICA procedure.
Else, the condition of colony has greater cost value than the imperialist, then the process skip the exchange operation and continues the same procedure until the process terminates.