Fog computing-based beam allocation and co-operative task distribution model for green 5G mobile network

Abstract

The communication and computation world is being gone through a lack of proper power consumption models in multi-tier small cell-based heterogeneous networks. Efficient spectrum utilization with proper task distribution is an emerging criterion for 5G-based wireless communication. Delay incurred by task offloading degrades the quality of services (QoS). To reduce power consumption, delay, and increase spectral efficiency, fog-based power-efficient beam allocation and task distribution model for IoT network are proposed. Femtolet acts as a base station and performs also computational assignments like a cloudlet. Hence, femtolet can act as a fog device. Femtolet allocates beam to each IoT device based on maximum spectral efficiency. Secondly, the task distribution is performed. If the femtolet itself can perform tasks, it processes them and sends back to the IoT device. The global femtolet concept appears when two or more femtolets are needed to perform tasks. The connection between IoT devices (IoTD) and femtolet is done by 5G beamforming. Mixed integer linear programming is formulated. Power consumption, delay, and spectral efficiency are the QoS factors of the IoT network. The simulation result demonstrates that proposed architecture decreases power consumption and delay by 23–47% and 15–25%, respectively, than cloud-sensitive approaches. The proposed approach enhances spectral efficiency and SINR by 12–15% and 15–25%, respectively. The comparative analysis with existing approaches shows that this model is novel, green, and fast.

Introduction

In the area of advanced mobile communication, power, delay, and spectrum optimization are the major parameters when QoS of 5G mobile networks is concerned [1, 2]. The communication between mobile devices and fog [3] or cloud system [4] endures with massive transmission power consumption and a lack of spectral efficiency problems, especially when using 4G/LTE-A [5] connections. Spectrums are in underutilized scenario due to interference in the network and also lack of proper utilization of spectrums [6]. That increases the huge power consumption of the network. Spectrum allocation in mobile networks deals with available and unutilized frequencies across the base stations to minimize the interference between base stations. This problem can be formulated as an integer linear program in which binary variables indicate whether a frequency is assigned to an antenna. In some cases, variables are not discrete. Mixed Integer Linear Programming (MILP) [7] concept comes in the scenario to solve the spectrum allocation problem. In this article, femtolet-beamforming is used. In the fifth-generation (5G) network, different technologies [8] can be used to overcome the above crisis. Beamforming is one of them [9]. Beamforming allows a proficient organization of intercell interference among mobile base stations [10]. This technique is very much efficient for the indoor region as short-distance data transmission occurs. The beamforming technique can be divided into two modes: (i) fixed beamforming mode; (ii) desired signal maximization and interference minimization mode. The second one is also known as adaptive beamforming [11]. In this proposed article, we have used an adaptive beamforming approach. Multimedia applications are the most common mobile application, whereas these applications need extensive power consumption and more computing resources. Communication along with computation is the desired phenomenon of the modern wireless industry. The cooperation between base stations is elementary to allocate radio access and computing requests in a most effective manner. To address these challenges, the femtolet is discussed and has considered a fog device [12]. Femtolet is an energy efficient and advanced delay-sensitive device for the indoor zone [13]. The mix of femtocell and cloudlet is represented as femtolet. Femtocell is a little cell home base station, and cloudlet plays out the computational assignment as a small cloud environment. It gives the administrations of femtocell just as cloudlets at minimum latency. Using a femtolet as a co-operative fog device-related tasks can be performed.

Motivations and contributions

One of the challenging issues in the 5G wireless network is to connect a massive number of devices into the network and execute the total task within a limited period for maintaining the QoS of the network. To connect more devices, effective spectrum allocation is a critical problem. The research incentive is to provide the power spectrum and delay optimized fog-based IoT network. Power, spectrum, and delay are important parameters for QoS assessment in heterogeneous 5G mobile networks [14]. For such circumstances, the femtolet device comes into the scenario. We aim to use a femtolet to serve each IoT device with proper spectrum using beamforming technology and also execute the assigned task with minimum delay. In this article, femtolet and distant cloud server act in a proficient way that in maximum scenarios task can be performed by a femtolet or a group of femtolet. If none of the femtolet can perform the task fully, then the task or partial task is transferred to the cloud server for concluding the total task. Main contributions of this article are as follows:

1. A fog-based power efficient beam allocation and task distribution model for a green IoT network is proposed using 5G beamforming to meet our objectives.

2. In the proposed model, IoTDs are registered under a fog device using beamforming. Femtolet provides short-distance communication and hence beamforming can be used. Femtolet allocates beam to the IoTDs based on maximum spectral efficiency.

3. After assigning the proper spectrum to each IoTDs, IoTD requests a fog device to execute tasks. If a femtolet cannot perform the task individually, it requests a neighboring femtolet That femtolet can be a single femtolet or a group of femtolet to finish the total task.

4. The power consumption, signal-to-interference plus noise ratio (SINR), spectral efficiency, and delay for the proposed approach are calculated and compared with other fog and cloud-based approaches to show that our proposed model balances power consumption and utilize proper spectrum. Experimental results obtained from the university laboratory define the performance of the proposed model.

The remaining article is structured as follows: Sect. 2 discusses the literature survey of the proposed scenario. Section 3 describes the proposed model architecture. The power consumption and delay models are illustrated in Sect. 4, the performance of the model is analyzed in Sect. 5, and Sect. 6 concludes the article.

Related work

Pricing strategy-based energy-efficient cloud computing has been discussed in [15]. Multiple cloud service providers compete with each other to increase their profit. The customers are related to service-oriented and offered price value to cloud service providers for the better quality of the service. In [15], the author has discussed the bi-level optimization strategy; level one is the slow timescale optimization procedure (STOP) for economic efficiency, and another one is fast timescale optimization procedure (FTOP) for energy-efficient job scheduling. For better spectrum availability in a small region, small cell-like femtocell plays an important role [16]. In indoor and border areas, femtocells provide better signal strength than other base stations. Power consumption of small cell-like femtocell base stations is less compared to other base stations and the coverage area of a femtocell is 10–20 m. It increases the data rate in the indoor region. Cloudlet-based resource allocation and task scheduling in mobile cloud computing have been discussed in [17]. Cloudlet has enough capacity to performs multiple mobile instruction, but it is less sufficient than traditional large capacity cloud computing. In [17], the authors discuss task scheduling and load aware resource allocation depending on the different types of users and their quality of service requirements. An intensive mobile application has been growing in the next-generation mobile cloud computing paradigm [18]. To serve these massive applications, cloudlet is a good solution for wireless communication. The author discusses multiple cloudlet-based mobile cloud computing system for reducing the power consumption and the delay for the intensive mobile applications. In [19], the authors propose cloudlet-assisted-based energy and delay optimization for offloading the computation problem in mobile cloud computing. The mobile device connects to the cloudlet through a small base station and if a small base station is not available under the macro base station, it connects to the cloud server with the macro base station. Application-aware cloudlet selection for mobile cloud computing has been discussed in [20]. The authors discuss the multi-cloudlet environment for different types of applications and their processes. From multiple cloudlets, the mobile device selects the nearest cloudlet to reduce the latency. In [12], the authors proposed an energy-efficient network device called “Femtolet,” it is the combination of a femtocell base station and the cloudlet device. Femtolet consumes less power consumption as the same as femtocell base station and it performs low latency computation for data instruction. Femtolet is easy to deploy and operate like plug and play manner. Small cell-based machine to machine traffic scheduling over long-term evaluation (LTE) is illustrated in [21]. In this article, a multi-objective optimization based on MILP for throughput minimization is discussed. Another MILP model-based power optimization problem is described in [22]. Here, cloudlet-based mobile cloud computing (MCC) [23] environment is considered. Mobile edge computing comes into the scenario to reduce the delay of a network. A mobile edge computing-based cost-effective cloudlet selection strategy is proposed. To measure the quality of service (QoS) of a network, cloud system-based architecture is discussed in [24]. The 5G technology comprises all types of advanced characteristics which makes 5G technologies the most powerful and in enormous requirement soon. Beamforming is another emerging technology of the 5G mobile network. Theoretical feasibility of mmWave beamforming is described in [25] for 5G cellular communication. MIMO and beamforming solutions for 5G are illustrated in [26].

In this article, the femtolet-beamforming model is considered for spectrum and power efficiency. Femtolet here is considered as a fog device. Communication between an IoTD and a femtolet is done using beamforming technology.

Proposed fog computing-based power efficient beam allocation and task distribution model for 5G network

IoTDs in the proposed model are connected with the femtolet with using beamforming. The fog devices are connected co-operatively as shown in Fig. 1.

[See PDF for image]

Fig. 1

Proposed fog computing-based beam allocation and task distribution model for green 5G network

Mathematical description for beam selection

We consider the total M number of beams which are transmitted from the femtolet.

For all beam, A, allocated to IoT devices is defined as

1

$$B_A=P_{A,i,n}·{\rho }_{A,i,n}·\overrightarrow{A}$$where $P_{A,i,n}$ denotes the power consumption for user i on subcarrier n for a beam A. ${\rho }_{A,i,n}$ denotes the frequency allocated for user i on subcarrier n for a beam A and $\overrightarrow{A}$ denotes the direction of the vector A for user i on subcarrier n.

For all the beam B allocated from femtolet to mobile devices is defined as

2

$$B_B=P_{B,i,n}·{\rho }_{B,i,n}·\overrightarrow{B}$$where $P_{B,i,n}$ denotes the power consumption for user i on subcarrier n for a beam B. ${\rho }_{B,i,n}$ denotes the frequency allocated for user i on subcarrier n for a beam B and $\overrightarrow{B}$ denotes the direction of the vector B for user i on subcarrier n.

For all the beam M allocated from femtolet to mobile devices is defined as

3

$$B_M=P_{M,i,n}·{\rho }_{M,i,n}·\overrightarrow{M}$$where $P_{M,i,n}$ denotes the power consumption for user i on subcarrier n for a beam M. ${\rho }_{M,i,n}$ denotes the frequency allocated for user i on subcarrier n for a beam M and $\overrightarrow{M}$ denotes the direction of the vector M for user i on subcarrier n.

then the total global femtolet beams are

4

$$\begin{array}{cc}\hfill B_T& =B_A+B_B+\cdots +B_M=P{}_{A,i,n}·{\rho }_{A,i,n}·\overrightarrow{A}\hfill \\ \hfill & +P_{B,i,n}·{\rho }_{B,i,n}·\overrightarrow{B}+\cdots +P_{M,i,n}·{\rho }_{M,i,n}·\overrightarrow{M}\hfill \\ \hfill & =\sum _{j=A}^M{P}_{j,i,n}+\sum _{j=A}^M\overrightarrow{j}\hfill \\ \hfill \end{array}$$

Fog devices are connected with the remote cloud server whenever it is required to finish a task partially or fully. The proposed method is described in Algorithm 1. Femtolet-mixed integer linear programming is discussed in Table 1. A group of femtolets integrates for creating a global femtolet to finish the task partially or fully. As all the local femtolets are mixed and create a single global femtolet, fog devices reside in the same layer so it is a linear approach also. Femtolets are chosen depending on the value of the available beam. Thus, we can say that femtolet-mixed integer linear programming is justified for our proposed network scenario.

Table 1. Description of femtolet mixed-integer linear programming

Femtolet	A group of local femtolets integrates to create a global femtolet
Mixed	We consider some small local number of femtocells placed in a cluster. All the local femtolets are mixed and create a single global femtolet
Integer	Each local femtolet has a fixed integer capacity to operate an instruction. When all the local femtolets (LF) are merged and create a global femtolet (GF), the capacity will also be merged and create an integer value. $G{F}_{\mathit{capacity}}={\sum }_{i=1}^n{(L{F}_{\mathit{capacity}})}_i$, where n number of LF is considered and the summation result of total LF is the GF capacity
Linear	All the femtolets are placed in the same layer or middle layer in the proposed architecture
Programming	The femtolets are selected based on the value of the available free beam

Femtolet has been used in the proposed approach as a fog device in the middle layer of the network as shown in Fig. 1. We have studied three cases for the proposed model as follows (Fig. 2).

[See PDF for image]

Fig. 2

Case studies for proposed fog computing-based beam allocation and task distribution model for green 5G network

In the proposed, fog-based network IoTDs are registered under a fog device as shown in Fig. 1. When an IoTD has to execute a task, it requests a femtolet. Femtolet first calculates the maximum spectral efficiency for each beam which is assigned to the IoT device. Then, IoTD requests for a task to perform to the femtolet. Femtolet checks if it can serve the purpose individually or it needs another or a group of femtolet. The task is performed either by a femtolet or a combination of a femtolet and a remote cloud server.

Mathematical model of the proposed fog computing-based beam allocation and task distribution model for green 5G network

Power consumption model for the proposed approach

Power consumption for Case 1 scenario

Here, the task is sent to the femtolet from the IoTD. Thus, the power consumption is calculated as follows.

Uplink power consumption to the fog device femtolet from the IoTD

5

$${\text{Pow}}_{\text{up}\_1}=P_{\text{send}}(D_{\text{uplink}}/R_{\text{uplink}})$$

Downlink power consumption to the fog device femtolet from the IoTD

6

$${\text{Pow}}_{\text{down}\_1}=P_{\text{receive}}(D_{\text{downlink}}/R_{\text{downlink}})$$

Power consumption for executing the task

7

$${\text{Pow}}_{\text{exe}\_1}=P_{\text{femtolet}}(I/S_{\text{femtolet}})$$

Total power consumption of case 1 is given as

8

$${\text{Pow}}_{\text{Tot}\_1}={\text{Pow}}_{\text{up}\_1}+{\text{Pow}}_{\text{down}\_1}+{\text{Pow}}_{\text{exe}\_1}$$

Power consumption for Case 2 scenario

Here, the IoTD requests the fog device, and the task is offloaded to the nearby femtolet of the femtolet. Therefore the power consumption from the IoTD to the femtolet, and the femtolet to the nearby femtolet is calculated as

Uplink power consumption

9

$${\text{Pow}}_{\text{up}\_2}=P_{\text{send}}(D_{\text{uplink}}/R_{\text{uplink}})$$

Downlink power consumption

10

$${\text{Pow}}_{\text{down}\_2}=P_{\text{receive}}(D_{\text{downlink}}/R_{\text{downlink}})$$

Power consumption for executing

11

$${\text{Pow}}_{\text{exe}\_\text{femtolet}\_2}=P_{\text{femtolet}}(I_{\text{compute}}/S_{\text{femtolet}})$$

Power consumption for sending from one femtolet to another femtolet

12

$${\text{Pow}}_{\text{Fem}\_\text{Fem}\_\text{sen}\_2}=(D_{\text{transmit}}/R_{\text{transmit}})$$

Power consumption for receiving from one femtolet to another femtolet

13

$${\text{Pow}}_{\text{Fem}\_\text{Fem}\_\text{rec}\_2}=(D_{\text{receive}}/R_{\text{receive}})$$

Total power consumption in case 2

14

$$\begin{array}{cc}\hfill {\text{Pow}}_{\text{Tot}\_2}& =({\text{Pow}}_{\text{up}\_2}+{\text{Pow}}_{\text{down}\_2}+{\text{Pow}}_{\text{ex}\_\text{femtolet}\_2}\hfill \\ \hfill & +{\text{Pow}}_{\text{fem}\_\text{fem}\_2}+{\text{Pow}}_{\text{fem}\_\text{fem}\_\text{refc}\_2})\hfill \\ \hfill \end{array}$$

Power consumption for Case 3 scenario

Here, the task is offloaded from the IoTD to the cloud server through middle layer femtolet. After execution, the IoTD receives result via the femtolet. The power calculation for case 3 is as follows:

Uplink power consumption

15

$${\text{Pow}}_{\text{up}\_3}=P_{\text{send}}(D_{\text{uplink}}/R_{\text{uplink}})$$

Downlink power consumption

16

$${\text{Pow}}_{\text{down}\_3}=P_{\text{receive}}(D_{\text{downlink}}/R_{\text{downlink}})$$

Power consumption for executing in cloud

17

$${\text{Pow}}_{\text{exe}\_\text{cloud}\_3}=P_{\text{cloud}}(I_{\text{compute}}/S_{\text{femtolet}})$$

Power consumption for sending from femtolet to cloud

18

$${\text{Pow}}_{\text{Fem}\_\text{cloud}\_3}=P_{\text{send}}(D_{\text{transmit}}/R_{\text{transmit}})$$

Power consumption for receiving from cloud to femtolet

19

$${\text{Pow}}_{\text{cloud}\_\text{Fem}\_3}=P{}_{\text{receive}}(D_{\text{receive}}/R_{\text{receive}})$$

Total power consumption in case 3

20

$$\begin{array}{cc}\hfill {\text{Pow}}_{\text{Tot}\_3}& =({\text{Pow}}_{\text{up}\_3}+{\text{Pow}}_{\text{down}\_3}+{\text{Pow}}_{\text{exe}\_\text{cloud}\_3}\hfill \\ \hfill & +{\text{Pow}}_{\text{fem}\_\text{cloud}\_3}+{\text{Pow}}_{\text{cloud}\_\text{fem}\_3})\hfill \\ \hfill \end{array}$$

Delay calculation in offloading the tasks of the proposed approach

Delay calculation for Case 1 scenario

In this case, the task is sent to the femtolet from the IoTD. Thus, the propagation delay is collected as,

21

$$T_{p1}=\frac{D_{\mathit{mf}}}{S_p}$$where $D_{\mathit{mf}}$ is the distance between a fog device and IoTD. $S_p$ denotes the propagation speed. The IoTD communicates with the femtolet, in this case, using beamforming technology of the 5G network. The failure rate of the uplink and downlink delay is given below,

22

$$T_{cu1}=\frac{A_{\mathit{umf}}}{U_{\mathit{dmf}}}+F_{\mathit{umf}}·\frac{A_{\mathit{umf}}}{U_{\mathit{dmf}}}$$

23

$$T_{cd1}=\frac{A_{\mathit{dmf}}}{D_{\mathit{dmf}}}+F_{\mathit{dmf}}·\frac{A_{\mathit{dmf}}}{D_{\mathit{dmf}}}$$where $A_{\mathit{dmf}}$ is downlink data transfer quantity among IoTD and fog device femtolet. $D_{\mathit{dmf}}$ and $F_{\mathit{dmf}}$ denote downlink data transfer rate among IoTD fog device femtolet and failure rate in downlink among IoTD and femtolet, respectively. Delay for executing the task is calculated as,

24

$$T_{e1}=\frac{I}{S_f}$$where $S_f$ is the speed of the femtolet.

The total delay for case 1 is given as,

25

$$T_{case1}=T_{p1}+T_{cu1}+T_{cd1}+T_{e1}$$

Delay calculation for Case 2 scenario

Here, IoTD requests the fog device femtolet, and the task is offloaded to the nearby femtolet. Therefore the propagation arises from the IoTD to the femtolet and the femtolet to the nearby femtolet. Thus, the propagation delay is given as,

26

$$T_{p2}=\frac{D_{\mathit{mf}}}{S_p}+\frac{D_{\mathit{fp}}}{S_p}$$$D_{\mathit{fp}}$ is the distance between fog device femtolet and its adjacent femtolet. The IoTD connects with the nearby femtolet via the previous femtolet. Uplink delay and downlink delay for case 2 are given in equation in (23) and (24), respectively,

27

$$T_{cu2}=\left(\frac{A_{\mathit{umf}}}{U_{\mathit{dmf}}}+F_{\mathit{umf}}·\frac{A_{\mathit{umf}}}{U_{\mathit{dmf}}}\right)+\left(\frac{A_{\mathit{ufp}}}{U_{\mathit{dfp}}}+F_{\mathit{ufp}}·\frac{A_{\mathit{ufp}}}{U_{\mathit{dfp}}}\right)$$

28

$$T_{cd2}=\left(\frac{A_{\mathit{dmf}}}{D_{\mathit{dmf}}}+F_{\mathit{dmf}}·\frac{A_{\mathit{dmf}}}{D_{\mathit{dmf}}}\right)+\left(\frac{A_{\mathit{dfp}}}{D_{\mathit{dfp}}}+F_{\mathit{dfp}}·\frac{A_{\mathit{dfp}}}{D_{\mathit{dfp}}}\right)$$$A_{\mathit{umf}}$ and $U_{\mathit{dmf}}$ are the amount of data transferred in uplink and downlink data transfer between fog device femtolet and its nearby femtolet, respectively. $F_{\mathit{dfp}}$ is the failure rate.

As the speed of the femtolet is considered, delay required for executing the task is calculated as,

29

$$T_{e2}=\frac{I}{S_{\text{femtolet}}}$$

The total delay for case 2 is given as,

30

$$T_{case2}=T_{p2}+T_{cu2}+T_{cd2}+T_{e2}$$

Delay calculation for Case 3 scenario

Here, the task is offloaded from the IoTD to the cloud server through the fog layer. Let the time stamp of sending request to the neighboring femtolet is T_sen and the time stamp of receiving declination is T_dec. Then, the delay is (T_dec–T_sen). In this case, propagation arises from IoTD to femtolet, and femtolet to the cloud server. The propagation delay is calculated as,

31

$$T_{p3}=\frac{D_{\mathit{mf}}}{S_p}+\frac{D_{\mathit{fc}}}{S_p}$$$D_{\mathit{mf}}$ is the distance between IoTD and fog device femtolet. The IoTD connects remote cloud server via fog device femtolet. Uplink delay and downlink delay for case 3 are given in equation in (32) and (33), respectively,

32

$$T_{cu3}=\left(\frac{A_{\mathit{umf}}}{U_{\mathit{dmf}}}+F_{\mathit{umf}}·\frac{A_{\mathit{umf}}}{U_{\mathit{dmf}}}\right)+\left(\frac{A_{\mathit{ufc}}}{U_{\mathit{dfc}}}+F_{\mathit{ufc}}·\frac{A_{\mathit{ufc}}}{U_{\mathit{dfc}}}\right)$$

33

$$T_{cd3}=\left(\frac{A_{\mathit{dmf}}}{D_{\mathit{dmf}}}+F_{\mathit{dmf}}·\frac{A_{\mathit{dmf}}}{D_{\mathit{dmf}}}\right)+\left(\frac{A_{\mathit{dfc}}}{D_{\mathit{dfc}}}+F_{\mathit{dfc}}·\frac{A_{\mathit{dfc}}}{D_{\mathit{dfc}}}\right)$$$A_{\mathit{dmf}}$ and $D_{\mathit{dmf}}$ are the amount of data transferred in uplink and downlink data transfer between fog device femtolet and remote cloud server, respectively. $F_{\mathit{dfc}}$ is the failure rate.

As the speed of cloud server is considered, delay required for executing the task is calculated as,

34

$$T_{e3}=\frac{I}{S_{\text{cloud}}}$$

The total delay for case 3 is given as,

35

$$T_{case3}=T_{p3}+T_{cu3}+T_{cd3}+T_{e3}+(T_{\mathit{dec}}-T_{\mathit{sen}})$$

Let, p₁, p_2, and p₃ be the probability of case 1, case 2 and case 3, then the total delay for the network is calculated as,

36

$$T=p_1·T_{case1}+p_2·T_{case2}+p_3·T_{case3}$$

SINR of the proposed approach

The SINR for large base station device p on subcarrier s is

37

$${\text{SINR}}_{c,q,t}=\frac{{\text{Pow}}_{\text{Tot}\_3}{G}_{q,C,t}}{N_0\mathrm{\Delta }f+{\sum }_N{P}_m{G}_{q,t}+{\sum }_{N_{\mathit{sc}}}{\text{Pow}}_{\text{Tot}\_3}{G}_{q,C,t}}$$where power consumption by the remote cloud server to send the data for IoTD via fog device p on subcarrier s is P_c.

The SINR IoTD to fog device q on subcarrier t using beamforming technology is

38

$${\text{SINR}}_{sc,q,t}=\frac{{\left|{\text{Pow}}_{\text{Tot}\_2}{G}_{q,SC,t}\right|}^2}{N_0\mathrm{\Delta }f+{\sum }_N{\left|P_m{G}_{q,t}\right|}^2+{\sum }_{N_{\mathit{sc}}}{\left|{\text{Pow}}_{\text{Tot}\_2}{G}_{q,SC,t}\right|}^2}$$where power consumption by femtolet to send the data for IoTD z on subcarrier y is P_SC.

Spectral efficiency of the proposed approach

The spectral efficiency for IoTD q on subcarrier y t with SINR_SC,q,t

39

$${\text{SE}}_{SC,q,t}={log}_2(1+{\text{SINR}}_{SC,q,t})$$

Performance analyses

Analytical evaluation

This model proposes a three-layer network architecture. At the top of the layer, a remote cloud server resides. At the bottom, IoTDs take place. Between these two layers, fog computing occurs. The connection between a fog device and an IoTD is carried out using novel beamforming technology of 5G. For analytical evaluation, MATLABR2015 has been used.

Power consumption

The power transmitted by IoTDs in the proposed network is presented in Fig. 3. The result of the proposed approach is compared with conventional heterogeneous fog and cloud-based networks. Figure 6 shows that the proposed network is approximately 23–47% power-efficient than conventional heterogeneous fog and cloud-based networks.

[See PDF for image]

Fig. 3

Comparison of power consumption between the proposed network and conventional networks

Signal to interference plus noise ratio

The SINR of the proposed approach-based network and conventional heterogeneous fog, cloud network is calculated using Eqs. (37, 38). Figure 4 demonstrates that the proposed network increases SINR by approximately 15–25% than conventional heterogeneous fog, cloud networks.

[See PDF for image]

Fig. 4

Comparison of SINR between the proposed network and conventional networks

Spectral efficiency

The spectral efficiency of the proposed network and conventional heterogeneous fog, cloud network is calculated. Figure 5 illustrates that the proposed network approximately 12–15% more spectrally efficient than conventional heterogeneous fog, cloud networks.

[See PDF for image]

Fig. 5

Comparison of spectral efficiency between the proposed network and conventional networks

Delay of the proposed network

Delay of the proposed network and conventional heterogeneous fog, cloud network is calculated using Eq. (36). Figure 6 illustrates that the proposed network decreases delay approximately 15–25% than conventional heterogeneous fog, cloud network.

[See PDF for image]

Fig. 6

Comparison of delay between the proposed network and conventional networks

Experimental evaluation

Agilent EXG Vector Signal Generator (VSG) N5172B (9 kHz–13.6 GHz), Vector Signal Analyzer (VSA) [27] N9010A (10 Hz–13.6 GHz) and signal studio software are used for the experimental purpose which is present in our laboratory, as illustrated in Fig. 7. The characteristics of the transmission signal of the fog device femtolet. The femtolet can work as a home node base station as well as fog devices. The power level of the signal 3.76 dBm in the proposed approach is shown in Fig. 7.

[See PDF for image]

Fig. 7

Constellation diagram and frame summary for the generated signal of the proposed network

The constellation diagram for the proposed generated signal is shown in Fig. 7a. This diagram shows the modulated signal. Resolution bandwidth (Res BW) denotes as a bandpass filter in an intermediate frequency (IF) path and video bandwidth (VBW) filters noise. Res BW is set to 15 kHz. Figure 7b shows the frame summary where the different channels with their power level are presented. The signal at frequency level 4.0064 is generated using VSG and signal studio software and analyzed the signal using VSA shown in Fig. 8. Here, the reference bandwidth is (Res BW) 150 kHz and video bandwidth is 150 kHz.

[See PDF for image]

Fig. 8

The power level of spectrum for the proposed network

Comparison of the proposed approach with existing schemes

Comparison of the proposed scheme with existing cloud-based approaches [7, 9, 14, 21] is shown in Table 2.

Table 2. Comparison between the proposed and the existing schemes [7, 9, 14, 21]

Properties	Energy consumption in mobile cloud computing paradigm [21]	A delay-aware power optimization model for mobile edge computing systems [7]	Resource provisioning and scheduling in clouds [14]	Optimal transmit beamforming for two-tire HetNet [9]	Proposed approach: Proposed fog computing-based beam allocation and task distribution model for green 5G network
Working principle	Cloud Computing platform is considered	An edge Computing platform is considered	QoS perspective resource provisioning and scheduling	Beamforming design for simultaneous wireless information and power transfer (SWIPT)	Fog Computing platform is considered along with 5G beamforming
Use of small cells	×	×	×	√	√
Consideration of Indoor region	×	√	×	×	√
SINR is calculated	×	×	×	√	√
Spectral efficiency is calculated	×	×	×	×	√
Reduction in power than existing schemes in	–	–	–	–	~ 23 to 47%
Reduction in delay than existing schemes in	–	–	–	–	~ 15 to 25%
Remarks	In [7, 9, 14, 21] cloud computing paradigm or edge computing scenario is used which consumes low power. But in our approach, we have used fog computing environment with beamforming which consumes less power than any other cloud or edge computing paradigm. It is observed that the proposed approach reduces power consumption 23–47% and also decreases delay 15–25% than other approaches [7, 9, 14, 21]. Thus, it is a novel, fast, and green approach

Conclusion

This article proposes a fog computing-based power-efficient beam allocation and task distribution model for green 5G. When an IoTD has to perform a task, it requests the fog device femtolet using beamforming technology. Femtolet allocates the beam to each IoT device based on maximum spectral efficiency. Femtolet executes the task and returns the result to the IoTD. Femtolet forwards the request to a nearby fog device when it is unable to process the request independently. Then, it takes help from another nearby femtolet and forms a global femtolet. The nearby femtolet accomplishes the task and sends the result to the femtolet. The femtolet sends the result to the IoTD. In this article delay, power consumption, SINR, and spectral efficiency of the proposed strategy are calculated. Those are considered as the QoS factors of the network. The performance evaluation of the current scenarios is also experimentally demonstrated concerning the existing scenarios. Simulation results show that using the proposed scenario power consumption and delay of the network is reduced. Our proposed approach reduces the delay and power consumption by 15–25% and 23–27%, respectively, than existing models. This model increases SINR and spectral efficiency by 15–25% and 12–15%, respectively. Thus, we conclude that the proposed strategy is a power-efficient, delay and spectrum optimized promising network model for 5G communication.

Acknowledgements

The authors are thankful to the Department of Science and Technology (DST)-FIST for SR/FST/ETI-296/2011 and TEQIP-III.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Huynh, DT; Wang, X; Duong, TQ; Vo, NS; Chen, M. Social-aware energy efficiency optimization for device-to-device communications in 5G networks. Comput Commun; 2018; 120, pp. 102-111. [DOI: https://dx.doi.org/10.1016/j.comcom.2018.02.008]

2. Marsch, P; Da Silva, I; Bulakci, O; Tesanovic, M; El Ayoubi, SE; Rosowski, T; Kaloxylos, A; Boldi, M. 5G radio access network architecture: Design guidelines and key considerations. IEEE Commun Mag; 2016; 54, 11 pp. 24-32. [DOI: https://dx.doi.org/10.1109/MCOM.2016.1600147CM]

3. Kumari, A; Tanwar, S; Tyagi, S; Kumar, N; Obaidat, MS; Rodrigues, JJ. Fog computing for smart grid systems in the 5G environment: Challenges and solutions. IEEE Wirel Commun; 2019; 26, 3 pp. 47-53. [DOI: https://dx.doi.org/10.1109/MWC.2019.1800356]

4. Lavanya M, Shanthi B, Saravanan S (2020) Multi objective task scheduling algorithm based on SLA and processing time suitable for cloud environment. Comput Commun

5. Bulakci, Ö; Redana, S; Raaf, B; Hämäläinen, J. Impact of power control optimization on the system performance of relay based LTE-Advanced heterogeneous networks. J Commun Netw; 2011; 13, 4 pp. 345-359. [DOI: https://dx.doi.org/10.1109/JCN.2011.6157454]

6. Mukherjee, A; De, D; Deb, P. Interference management in macro-femtocell and micro-femtocell cluster-based long-term evaluation-advanced green mobile network. IET Commun; 2016; 10, 5 pp. 468-478. [DOI: https://dx.doi.org/10.1049/iet-com.2015.0982]

7. Jararweh, Y; Al-Ayyoub, M; Al-Quraan, M; L’oai, AT; Benkhelifa, E. Delay-aware power optimization model for mobile edge computing systems. Pers Ubiquitous Comput; 2017; 21, 6 pp. 1067-1077. [DOI: https://dx.doi.org/10.1007/s00779-017-1032-2]

8. Boccardi, F; Heath, RW; Lozano, A; Marzetta, TL; Popovski, P. Five disruptive technology directions for 5G. IEEE Commun Mag; 2014; 52, 2 pp. 74-80. [DOI: https://dx.doi.org/10.1109/MCOM.2014.6736746]

9. Li, B; Fei, Z; Chu, Z. Optimal transmit beamforming for secure SWIPT in a two-tier HetNet. IEEE Commun Lett; 2017; 21, 11 pp. 2476-2479. [DOI: https://dx.doi.org/10.1109/LCOMM.2017.2734759]

10. Huang, Y; Zheng, G; Bengtsson, M; Wong, KK; Yang, L; Ottersten, B. Distributed multicell beamforming with limited intercell coordination. IEEE Trans Signal Process; 2010; 59, 2 pp. 728-738.2815412 [DOI: https://dx.doi.org/10.1109/TSP.2010.2089621]

11. Kim HS, Ko H, Beh J, Lee T (2014) U.S. Patent No. 8,774,952. U.S. Patent and Trademark Office, Washington, DC

12. Mukherjee, A; De, D. Femtolet: a novel fifth generation network device for green mobile cloud computing. Simul Model Pract Theory; 2016; 62, pp. 68-87. [DOI: https://dx.doi.org/10.1016/j.simpat.2016.01.014]

13. Mukherjee, A; Deb, P; De, D; Buyya, R. C2OF2N: a low power cooperative code offloading method for femtolet-based fog network. J Supercomput; 2018; 74, 6 pp. 2412-2448. [DOI: https://dx.doi.org/10.1007/s11227-018-2269-x]

14. Hasan, M; Goraya, MS. Flexible fault tolerance in cloud through replicated cooperative resource group. Comput Commun; 2019; 145, pp. 176-192. [DOI: https://dx.doi.org/10.1016/j.comcom.2019.06.005]

15. Qiu, C; Shen, H; Chen, L. Towards green cloud computing: demand allocation and pricing policies for cloud service brokerage. IEEE Trans Big Data; 2018; 5, 2 pp. 238-251. [DOI: https://dx.doi.org/10.1109/TBDATA.2018.2823330]

16. Thakur, R; Mishra, S; Murthy, CSR. An energy and cost aware framework for cell selection and energy cooperation in rural and remote femtocell networks. IEEE Trans Green Commun Netw; 2017; 1, 4 pp. 423-433. [DOI: https://dx.doi.org/10.1109/TGCN.2017.2736007]

17. Liu, Y; Lee, MJ; Zheng, Y. Adaptive multi-resource allocation for cloudlet-based mobile cloud computing system. IEEE Trans Mob Comput; 2015; 15, 10 pp. 2398-2410. [DOI: https://dx.doi.org/10.1109/TMC.2015.2504091]

18. Tawalbeh, LA; Jararweh, Y; Ababneh, F; Dosari, F. Large scale cloudlets deployment for efficient mobile cloud computing. J Netw; 2015; 10, 1 70.

19. Liu, L; Guo, X; Chang, Z; Ristaniemi, T. Joint optimization of energy and delay for computation offloading in cloudlet-assisted mobile cloud computing. Wirel Netw; 2019; 25, 4 pp. 2027-2040. [DOI: https://dx.doi.org/10.1007/s11276-018-1794-0]

20. Roy, DG; De, D; Mukherjee, A; Buyya, R. Application-aware cloudlet selection for computation offloading in multi-cloudlet environment. J Supercomput; 2017; 73, 4 pp. 1672-1690. [DOI: https://dx.doi.org/10.1007/s11227-016-1872-y]

21. Peng, H; Wen, WS; Tseng, ML; Li, LL. Joint optimization method for task scheduling time and energy consumption in mobile cloud computing environment. Appl Soft Comput; 2019; 80, pp. 534-545. [DOI: https://dx.doi.org/10.1016/j.asoc.2019.04.027]

22. Abrignani, MD; Giupponi, L; Lodi, A; Verdone, R. Scheduling M2M traffic over LTE uplink of a dense small cell network. EURASIP J Wirel Commun Netw; 2018; 2018, 1 193. [DOI: https://dx.doi.org/10.1186/s13638-018-1206-2]

23. De, D. Mobile cloud computing: architectures, algorithms and applications; 2016; Cambridge, CRC Press: [DOI: https://dx.doi.org/10.1201/b19208]

24. Ghahramani, MH; Zhou, M; Hon, CT. Toward cloud computing QoS architecture: analysis of cloud systems and cloud services. IEEE/CAA J Automatica Sinica; 2017; 4, 1 pp. 6-18.3606684 [DOI: https://dx.doi.org/10.1109/JAS.2017.7510313]

25. Zhang, J; Ge, X; Li, Q; Guizani, M; Zhang, Y. 5G millimeter-wave antenna array: Design and challenges. IEEE Wirel Commun; 2016; 24, 2 pp. 106-112. [DOI: https://dx.doi.org/10.1109/MWC.2016.1400374RP]

26. Shi, Y; Zhang, J; Letaief, KB. Group sparse beamforming for green cloud-RAN. IEEE Trans Wirel Commun; 2014; 13, 5 pp. 2809-2823. [DOI: https://dx.doi.org/10.1109/TWC.2014.040214.131770]

27. Li CF, Hwang JK, Ma C, Lin CJ (2017) Software defined radio implementation of LTE R13 NB-IoT downlink vector signal generator. In: 2017 IEEE international conference on consumer electronics-Taiwan (ICCE-TW). IEEE, pp 69–70