Social-sine cosine algorithm-based cross layer resource allocation in wireless network

Received Feb 26, 2020 Revised Jul 22, 2020 Accepted Aug 13, 2020 Cross layer resource allocation in the wireless networks is approached traditionally either by communications networks or information theory. The major issue in networking is the allocation of limited resources from the users of network. In traditional layered network, the resource are allocated at medium access control (MAC) and the network layers uses the communication links in bit pipes for delivering the data at fixed rate with the occasional random errors. Hence, this paper presents the cross-layer resource allocation in wireless network based on the proposed social-sine cosine algorithm (SSCA). The proposed SSCA is designed by integrating social ski driver (SSD) and sine cosine algorithm (SCA). Also, for further refining the resource allocation scheme, the proposed SSCA uses the fitness based on energy and fairness in which max-min, hard-fairness, proportional fairness, mixed-bias and the maximum throughput is considered. Based on energy and fairness, the cross-layer optimization entity makes the decision on resource allocation to mitigate the sum rate of network. The performance of resource allocation based on proposed model is evaluated based on energy, throughput, and the fairness. The developed model achieves the maximal energy of 258213, maximal throughput of 3.703, and the maximal fairness of 0.868, respectively.


SYSTEM MODEL
This section presents the system model for cross layer resource allocation in CDMA-based wireless as-hoc network. Let us consider the energy constrained Cognitive Radio orthogonal frequency division multiple access (CR OFDMA) with M communicating pairs. Here, both the transmitter u and the receiver v are represented as   n K ,..., 2 , 1 : . If v u  , then the transmission system is considered as the time slotted OFDMA system at the particular time interval L  . Here, the slot synchronization is attained using beaconing approach. For every time slot, the particular time interval is given for achieving synchronization to perform spectrum detection and resource allocation. However, inter-carrier interference (ICI) produced by the frequency offset of side lobes pertaining to the transmitter u . In the physical layer, the frequency-based Rayleigh fading channel is considered for dividing whole spectrum into N subcarriers for guaranteeing every subcarrier by experiencing Rayleigh fading. The subcarrier set present in transmitter and receiver pair u is denoted as,   where, the term l u a refer to the data rate of th l subcarrier. The power allocated over th l subcarrier for th u transmitter is represented as l u g , and the term s u g signifies the receiving power.

PROPOSED RESOURCE ALLOCATION ALGORITHM
This section illustrates the proposed SSCA algorithm for resource allocation in wireless network. The cross-layer optimization is the combination of joint scheduling and resource allocation in wireless network along with medium access control (MAC), physical layer and the application layer are included in the unified cross-layer optimization. Here, the cross-layer optimization algorithm gets queue state information (QSI) and channel state interference (CSI) from MAC and physical layer. Consequently, the other input parameters, like energy and fairness, the cross-layer optimization entity makes the decision on resource allocation to maximize the sum rate of network. By varying the channel conditions, the cross-layer optimization entity updates the decision based on new input data. Here, the novel resource allocation strategy is developed using the proposed SSCA, which will be newly designed by integrating SSD [22] and SSA [23]. Also, for further refining the resource allocation scheme, the proposed SSCA uses the fitness based on energy [19] and fairness where the hard-fairness, proportional fairness, mixed-bias max-min, and maximum throughput is considered. The block diagram of the proposed SSCA-based resource allocation in the wireless network is shown in Figure 1.

Cross layer optimization
Cross-layer optimization model consists of different layers and different parameters. The three layers present in the cross-layer optimization. The adaptive modulation and coding (AMC) and CSI-reference signal (CSI-RS) feedback are available in physical layer, whereas the subcarrier assignment, adaptive power control, automatic repeat request (ARQ), forward error correction (FEC), and quality of service in the MAC layer along with the adaptive transmission rate in the network layer. In traditional OSI, the strict boundaries are present among the layers in which the data are provided in the given particular layer and every layer gives the independent solution with the own optimized adaptation, but it is very complex to fulfill all the requirements, like data rate, bit error rate, and the latency for various services. In the dynamic wireless networks, the QSI and CSI vary with the time, hence the network node adapts the reception and the transmission parameters to satisfy the power constraint and the QoS needs [5,24].

Proposed social-sine cosine algorithm for resource allocation
The proposed SSCA is the combination of SSD with SSA. The SSD algorithm is inspired by various evolutionary optimization approaches to minimize the SVMs parameters with the aim to enhance the system performance. The main aim of SSD is to search in the space for optimal or near-optimal solutions. This method is very efficient for generating improved features to tackle multi-objective optimization issues. Moreover, the method can solve the highly non-linear problems with complex constraints. The SCA algorithm is the population-based optimization and there is no guarantee for finding the solution with single run. This approach is utilized for creating various random solutions and fluctuate them towards the best solution based on sine function. Here, various random and adaptive variables are combined in order to emphasize the exploitation and exploration of the search space. The merits of the algorithm are that the algorithm exhibits better convergence speed, minimal error, and minimal computational time. Hence, integrating SSD in SSA produces better solution with the improved system performance.
The developed SSCA for cross layer resource allocation is illustrated in this section. The proposed SSCA is the combination of SSD [22] with SSA [23], and thus attains the advantages of SSD in SSA. The SSD algorithm is inspired by various evolutionary optimization approaches to minimize the SVMs parameters with the aim to enhance the system performance. The main aim of SSD is to search in the space for optimal or near-optimal solutions. This method is very efficient for generating improved features to tackle multi-objective optimization issues. Moreover, the method can solve the highly non-linear problems with complex constraints. The SCA algorithm is the population-based optimization and there is no guarantee for finding the solution with single run. This approach is utilized for creating various random solutions and fluctuate them towards the best solution based on sine function. Here, various random and adaptive variables are combined in order to emphasize the exploitation and exploration of the search space. The merits of the algorithm are that the algorithm exhibits better convergence speed, minimal error, and minimal computational time. Hence, integrating SSD in SSA produces better solution with the improved system performance. The advantages of the proposed method are better convergence speed, minimal error, minimal computation time and it produces better solution with the improved system performance.

Solution encoding
The solution encoding is the representation that is identified with the developed model. Assume d number of transmitters and m number of subcarriers from which s optimal solution is chosen by the developed model such that s value ranging from b s   1 , respectively. Here, the solution requires for deciding which transmitter to be allocated in which subcarrier.

Objective modelling
The fitness function is evaluated in order to obtain the better result. The optimal solution is determined from the previous iteration as each solution to obtain the better location. The objective function of the developed approach is formulated in terms of energy, and fairness, and is expressed by, b. Hard-fairness: It is also termed as round robin (RR)-based resource allocation. It is utilized for allocating time or frequency between the potential candidates with respect to any metric. Simple time division multiplexing (TDM) is the best example for RR scheduler where every node is given as the time slot to transmit in the regular intervals. The expression for hard fairness is denoted as, where, the term S A indicates the maximum number of users who equally shared the resources.
c. Max-min fairness: In Max-min fairness, the less number of resources allocated to every node is increased. In other words, the gap is minimized among maximum and minimum number of assigned resources to each user.
where, the transmit power of relay helping the source u on subcarrier is denoted as ( ) , and the data rate is represented as u c .
d. Proportional fairness: It implements time-enabled fairness and provides the good tradeoff among fairness and network throughput with respect to max-min fairness. Here, the nodes with the lower data rate take more time than those with the higher data rates, which leads to reduced network throughput.
The proportional fairness is expressed as, where, the term  represents the characteristic which priority is assigned to 0   , and the proportionality e. Mixed-bias: Mixed-bias aims to allocate the portion of total capacity available at the node through strongly biased policy, and rests are employed as the fairer policy. The expression of mixed-bias is given by, where, the terms 1 p and 2 p represents the proportionality factor, Maximum throughput: It is concerned only with resources allocation for maximizing the throughput.
Here, the node that transmits more data, gets access to resources first to obtain high sum-throughput.

Algorithmic steps of the developed SSCA-based allocation
The proposed SSCA is designed by integrating SCA in SSD. Here, the update equation of SSD is modified using the update equation of SCA algorithm. The modification makes the solution update to be more efficient, and it further improves the convergence of the optimization algorithm. The steps followed in the developed algorithm are illustrated below: Step 1: Initialization: The first step of the proposed SSCA algorithm is the initialization of the position of search agents in which, the total number of agents are identified by the user. The location of agents is represented as, where, t X  represents the agents position at time t , and the number of variables are denoted as .
Step 2: Objective function evaluation: The fitness is calculated for each solution on the basis of fitness function depicted in (4). The fitness function is considered as the maximization function, and solution with maximum fitness is considered as the best solution. Step 3: Update the solution based on SSCA algorithm: After evaluating the objective function, the solution undergoes position update on the basis of SSCA. The update equation of SSD velocity t k  is expressed as,   (18) where, the term represents the location of current solution at ℎ iteration, refer to the target position, denotes the absolute value, and the random numbers are denoted as 1 , 2 , 3 are the random numbers, respectively. Substituting (18) in (13), where, 1 = − . Here, the term o denotes current iteration, Z refer to the maximal iterations, x represents the constant.
where, ∅ , , denotes the three best solutions, and represents the best solution.
The term refers to the ability of SSCA for better solution. The above equation specifies the updated equation of the proposed SSCA, which in turn used to perform cross layer resource allocation effectively.
Step 4: Compute the feasibility: After evaluating the updated position, the objective function of each solution is computed and the solution yielding maximum fitness is considered as best solution.
Step 5: Termination: The steps from 2 to 4 are repeated until the specified iteration met or the best solution is obtained. Algorithm 1 represents the pseudo code of developed SSCA algorithm.

RESULTS AND DISCUSSIONS
The analysis of cross-layer resource allocationusing the proposed SSCA-based allocation is elaborated in this section to prove the effectiveness of the proposed model.

Experimental set-up
The proposed method is executed in 4GB RAM, Windows 8 OS with Intel core i-3 processor and the implementation is done in MATLAB.

Evaluation metrics
The performance revealed by the developed approach is evaluated using energy, throughput and the fairness.

Comparative methods
The performance increased by the developed method is evaluated by comparing the proposed with existing methods, like waterfilling method [16], Security aware energy efficient allocation [19], distributed energy efficient allocation [24], respectively. Figure 2 shows the comparative of the developed SSCA-based allocation with respect to energy, throughput and the fairness with respect to users. Figure 2(a) represents the comparative analysis of the proposed SSCA-based allocation in terms of energy. When the count of user is 2, the energy obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 32943W, 33157W, and 31168W, while the proposed SSCA-based allocation obtained better energy of 32173, respectively. When the count of user is 6, the energy obtained by the proposed SSCA-based allocation is 38293W, while the percentage of improvement reported by the proposed method in comparison with the exiting waterfilling method, and security aware energy efficient allocation, distributed energy efficient allocation is 35678W, 37920W, and 33999W, respectively. Figure 2(b) represents the comparative analysis of the proposed SSCA-based allocation in terms of throughput by varying the users. When the count of user is 6, the throughput obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocationis 3.213mbps, 3.241mbps, and 3.247mbps, while the proposed SSCA-based allocation obtained better throughput of 3.292mbps, respectively. When the count of user is 8, the throughput obtained by the proposed SSCA-based allocation is 3.445mbps, while the percentage of improvement reported by the proposed method in comparison with the exiting waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 3.320mbps, 3.392mbps, and 3.414mbps, respectively. When the number of user is 10, the throughput obtained by the proposed SSCA-based allocation is 3.608mbps, while the percentage of improvement reported by the proposed method in comparison with the exiting waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 3.396mbps, 3.540mbps, and 3.577mbps, respectively. Figure 2(c) represents the comparative analysis of the proposed SSCA-based allocation in terms of fairness by varying the users. When the count of user is 6, the fairness obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 0.760, 0.770, and 0.788, while the proposed SSCA-based allocation obtained better fairness of 0.849, respectively. When the count of user is 8, the fairness obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 0.776, 0.855, and 0.860, while the proposed SSCAbased allocation obtained better fairness of 0.861, respectively. When the count of user is 10, the fairness obtained by the existing waterfilling method, security aware energy efficient allocation, and distributed energy efficient allocation is 0.805, 0.863, and 0.866, while the proposed SSCA-based allocation obtained better fairness of 0.868, respectively. Figure 3 shows the comparative of the developed SSCA-based allocation in terms of energy, throughput and the fairness with respect to transmitters. Figure 3(a) represents the comparative analysis of the proposed SSCA-based allocation in terms of energy. When the count of transmitter is 5, the energy obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 33336W, 31604W, and 31420W, while the proposed SSCA-based allocation obtained better energy of 31709W, respectively. When the count of transmitter is 20, the energy obtained by the proposed SSCA-based allocation is 39110W, while the percentage of improvement reported by the proposed method in comparison with the exiting waterfilling method, and Security aware energy efficient allocation, distributed energy efficient allocation is 35823W, 38662W, and 39557W, respectively. Figure 3(b) represents the comparative analysis of the proposed SSCA-based allocation in terms of throughput by varying the transmitters. When the count of transmitter is 10, the throughput obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 3.019mbps, 3.039mbps, and 3.134mbps, while the proposed SSCA-based allocation obtained better throughput of 3.218mbps, respectively. When the count of transmitter is 15, the throughput obtained by the proposed SSCA-based allocation is 3.265mbps, while the percentage of improvement reported by the proposed method in comparison with the exiting waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 3.097mbps, 3.157mbps, and 3.208mbps, respectively. When the number of transmitter is 25, the throughput obtained by the proposed SSCA-based allocation is 3.609 mbps, while the percentage of improvement reported by the proposed method in comparison with   Figure 4 illustrates the comparative of the developed SSCA-based allocation with respect to energy, throughput and the fairness with respect to subcarriers. Figure 4(a) represents the comparative analysis of the proposed SSCA-based allocation in terms of energy. When the count of subcarrier is 16, the energy obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 21200W, 18025W, and 18433W, while the proposed SSCA-based allocation obtained better energy of 22098W, respectively. When the count of subcarrier is 128, the energy obtained by the proposed SSCA-based allocation is 134377W, while the percentage of improvement reported by the proposed method in comparison with the exiting waterfilling method, and Security aware energy efficient allocation, distributed energy efficient allocation is 129883W, 125226W, and 127958W, respectively. Figure 4(b) represents the comparative analysis of the proposed SSCA-based allocation in terms of throughput by varying the subcarriers. When the count of subcarrier is 32, the throughput obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 2.327mbps, 2.489mbps, and 2.631mbps, while the proposed SSCA-based allocation obtained better throughput of 2.700mbps, respectively. When the count of subcarrier is 64, the throughput obtained by the proposed SSCA-based allocation is 3.171mbps, while the percentage of improvement reported by the proposed method in comparison with the exiting waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 2.852mbps, 2.898mbps, and 2.970mbps, respectively. When the number of subcarrier is 256, the throughput obtained by the proposed SSCA-based allocation is 3.703mbps, while the percentage of improvement reported by the proposed method in comparison with the exiting waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 3.486mbps, 3.502mbps, and 3.685mbps, respectively. Figure 4(c) represents the comparative analysis of the proposed SSCA-based allocation in terms of fairness by varying the subcarriers. When the count of subcarrier is 32, the fairness obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 0.735, 0.739, and 0.741, while the proposed SSCA-based allocation obtained better fairness of 0.743, respectively. When the count of subcarrier is 64, the fairness obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 0.738, 0.767, and 0.776, while the proposed SSCA-based allocation obtained better fairness of 0.809, respectively. When the count of subcarrier is 256, the fairness obtained by the existing waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation is 0.794, 0.820, and 0.823, while the proposed SSCA-based allocation obtained better fairness of 0.865, respectively.  Table 1 depicts the comparative discussion of developed method by varying the users, transmitters, and the subcarriers. The energy obtained by the existing methods, like waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation by varying the subcarriers is specified as 256553W, 254324W, and255357W, while the proposed SSCA-based allocation obtained better energy of 258213W, respectively. The throughput obtained by the existing methods, like waterfilling method, Security aware energy efficient allocation, and distributed energy efficient allocation by varying the subcarriers is specified as 3.486mbps, 3.502mbps, and 3.685mbps, while the proposed SSCA-based allocation obtained better throughput of 3.703mbps, respectively.