A novel distributed spectrum management in mobile edge computing based cognitive radio Internet of Things networks

Abstract

The integration of cognitive radio (CR) in Internet of Things (IoT) is an effective step into the smart technology world. The capability of CR can effectively solve spectrum-related issues for IoT applications, but this association is still a big challenge that has led to a new research dimension of CR-based IoT. To this extend, in this paper the authors propose a novel distributed spectrum management approach based on mobile edge computing (MEC) technology in cooperative environment that enables CRIoT devices to share the unutilized spectrum efficiently. The simulation results show that the proposed solution achieves good performance in terms of spectrum access/sharing and maintains a balance energy consumption of CRIoT users within lower latency.

Keywords

Internet of Things cognitive radio mobile edge computing MEC services spectrum management energy consumption mobility analysis process decision making

1. Introduction

The recent developments in information and communication technologies have presented a new paradigm: the Internet of Things (IoT). The IoT paradigm provides connectivity to objects for information exchange from one remote place to another via the Internet. These objects sense the physical world and harvest information about it, and send the harvested data to remote users via the Internet. It provides the accessibility of information from anywhere in anytime [14].

The evolution of IoT devices has generated huge demand for wireless bandwidth in order to meet the operational needs of new generation IoT applications. However, several IoT nodes do not own the bandwidth and therefore are called secondary users (SUs). While, a large portion of spectrum is under-utilized by licensed users, also known as primary users (PUs). Cognitive radio has received much attention of the research community as an important means for addressing the bandwidth needs of IoT applications [7].

So, CRNs is merged with the IoT technology to be named Cognitive Radio Internet of Things (CRIoT); that can effectively utilize the unused spectrum who already owned by PU. CR is an intelligent wireless communication technique, which realizes his surroundings in all cases. Joseph Motila is the first to propose CR. Using cognitive radio in order to make the best use of the available spectrum and increase productivity. It is considered as a smart radio that knows the spectrum state. It consists of a SU and a PU. The exploitation of the spectrum is done by SUs, provided that it does not interfere with the PUs [19].

However, one of the key issues in CR networks is to avoid device level collisions and interferences while maintaining efficient spectrum usage. A non-cooperative node can cause harmful interference to its neighbours and hence can reduce the overall spectrum usage [8]. The cooperation is a suitable technique for CRIoT, in which users cooperate with each other for spectrum sharing. Several CRIoT nodes can be deployed around each other and the idea is to exploit the capability of neighbouring CRIoT nodes to enhance the network performance. The cooperation between the PUs and the SUs is done by leasing the spectrum from the primary network to the SUs, in order to enhance the sum of utilities for both the PUs and SUs.

However, the existence of several PUs and CRs, and energy levels required for transmission create problems for CRs to sense and decide spectrum efficiently. This creates negative impact on the network [1]. Hence, if the integration of CR in IoT offers advantages then at the same time it poses new challenges which might be tackled by the promising technologies such as the mobile edge computing (MEC).

Unlike mobile cloud computing which has long latency since the powerful severs are always far away from the wireless devices. MEC integrates MEC servers with the access points or the base stations at the edge of mobile networks [2]. The CRIoT node can send their spectrum allocation demands to the MEC sever for efficient spectrum allocation and receive the results instantaneously.

This paper aims to solve the problem of spectrum scarcity caused by the growing demand for spectrum space of CRIoT and the problem of interference between different IoT devices which makes very difficult to allocate spectrum bands to all them. A distributed solution is preferable for a CR environment in order to meet the needs of CRIoT devices in terms of response time and low latency. A centralized solution is easy to set up but has shortcomings when the number of nodes is large where the performance will degrade and if the server crashes the whole system collapses. To address the above problems, the authors propose a distributed spectrum allocation/sharing approach based on MEC technology and cooperative behaviour between IoT devices in a CR network and cooperation between MEC servers. By bringing spectrum management mechanism to the MEC of cellular networks, a spectrum management services are put in place for determining the manipulation of each allocation request received from CRIoT devices and contributes to the conservation of their energy which is deemed to be spent on spectrum detection, computation and processing in order to share the appropriate spectrum to the application of IoT. Thus, the intention of this paper is to utilize the benefits of both mobile edge computing and cooperation between CRIoT devices to have a better use of the spectrum and better system performance and energy consumption.

The rest of the paper is organized as follows. The related work is discussed in Section 2. The system model is shown in Section 3. Section 4 illustrates a detailed explanation of the proposed approach, which is evaluated in Section 5. Finally conclusion of this work and future perspectives are stated in Section 6.

2. Related works

In literature, several researchers have addressed the spectrum allocation and sharing problem for CRIoT. This section will briefly cover some of these protocols.

In [16], the authors present a channel allocation strategy for IoT using opportunistic spectrum access. The proposed protocol for the distributed assignment of channels in cognitive Internet of Things networks is adopted in order to quickly respond to changes in topology and channels that may occur on the network. The main contribution of this work is in the use of the historical network traffic in conjunction with the ZAP algorithm (an algorithm for cognitive distributed assignment of channels) as a criterion for decision making in the allocation of channels. The performance evaluation shows that using the historical network traffic as a decision criterion leads to better results when compared with the simple use of the network topology as a criterion. The downside of this method is the complexity of the processes performed by each IoT, an IoT may even be responsible for assigning channels to others. This is not appreciated due to the limited capacity of the IoT.

The paper in [15] addresses the spectrum allocation problem with respect to both spectrum utilization and network throughput in the cognitive radio-based IoT. The authors propose a concurrent transmission model in the network which reveals the constraints of mutual interference and resource competition in links concurrent transmissions. The spectrum allocation problem is formulated as a multi-objective optimization problem and the spectrum allocation plan is transformed into the solution in genic algorithms. Then, the Non-dominated Sorting Genetic Algorithm-II is applied to solve the multi-objective spectrum allocation problem. Simulation results validate that the proposed strategy can search the optimal solutions efficiently and satisfy the requirements of spectrum allocation in various cases. The authors did not indicate the node or nodes responsible for launching their algorithm. The lack of cooperation between PUs and SUs can cause interference. In addition, the fairness and the quality of service (QoS) are not studied in this work. The proposed strategy has not been compared with other strategies existing in the literature. Another drawback of this solution is the scalability of the system as genetic algorithms generally take a long time, which allows the algorithm not to work in real time with a high number of CRIoTs.

In [17], a new cognitive radio spectrum sensing and sharing scheme is developed for IoT systems. In order to ensure cooperative spectrum sensing and sharing, the authors have added a new concept, called reciprocal fairness, and used a game theoretical tool in the proposed scheme. To reason the adaptive spectrum sensing and sharing problems in IoT systems, game theory is well-suited and an effective way. The experimental result illustrates that the game-based approach can get a better spectrum utilization, which can make the IoT performance maximal. The disadvantage of this strategy is the use of derived equations which are too complicated to be applied in real time, especially that the proposed algorithm works every time an empty location is detected.

The authors in [3] focus on a dynamic spectrum access strategy for IoT applications in two types of radio systems: cellular networks and cognitive radio-enabled low power wide area networks (CR-LPWANs). The aim is to maximize the spectrum capacity for the unlicensed users while ensuring that it never interferes with the licensed network. Therefore, in this paper, a dynamic spectrum access strategy for CR-LPWANs operating in both licensed and unlicensed bands is proposed. The results show that the proposed strategy can maximize the spectrum capacity for the unlicensed users using IoT applications as well as keep the service quality of the licensed users independent of them. In contrast, the proposed strategy is a centralized solution where the Base Station collects all information acquired by each of the IoT devices. The collected information would be sent to the data server. The detection of licensed bands is done by IoT devices, which increases its energy consumption. IoT nodes exploit the detected licensed bands to transmit in an opportunistic manner only when it does not interfere with any PUs, which is really very difficult to guarantee. Another problem is that centralized decision making at the server level generates high latency and when the number of IoTs is very large the server will be a bottleneck. In addition, the performance of the proposed protocol is not compared with any other protocol.

A channel assignment mechanism for the hardware-constrained CR-IoT networks under time-varying channel conditions with guard-band channels awareness is proposed in [11]. The objective of this assignment is to serve the largest possible number of CR-IoT devices by assigning the least number of idle channels to each device subject to rate demand and interference constraints. The proposed channel assignment in this paper is conducted on a per-block basis for the contending CR-IoT devices while considering the time-varying channel conditions for each CRIoT transmission over each idle channel, such that spectrum efficiency is improved. Specifically, the channel assignment problem is formulated as a binary linear programming problem, which is NP-hard. Thus, a polynomial-time solution is proposed using a sequential fixing algorithm that achieves a suboptimal solution. The simulation results demonstrate that the proposed assignment provides significant increase in the number of served IoT devices over existing assignment mechanisms. The downside of this solution is that the procedure for collecting spectrum information has not been addressed. In addition, the solution did not specify which node makes the spectrum sharing decision, i.e. in which device the algorithm is launched. Furthermore, the number of used IoT devices in simulation to evaluate the protocol is too small.

In [4], the authors have investigated spectrum leasing based on the idea that secondary users can earn spectrum access in exchange for cooperation with the primary user. This paper formulate the resource sharing problem between the primary user and secondary users as cooperative games, and use the Nash Bargaining Solution (NBS) and selective cooperation to achieve the optimal strategy. Analysis and numerical results show that spectrum leasing based on cooperative games will enhance the sum of utilities for both the PUs and the SUs. In the proposed solution, a PU selects one and only one SU to share its bandwidth with it, and therefore this solution does not guarantee satisfaction for all SU requests since the majority of the SU requests will be failed due to the lack of PUs. In addition, the performance of the proposed solution is not compared with any other strategy, also the strategy was evaluated for a very small number of SUs (only 4 SUs), a single PU and the authors did not study its scalability.

As for the use of fog/edge computing solutions in the cognitive radio network, the majority of his research has taken advantage of these technologies to solve the problems of limited computing and storage capacity in CRIoT devices. No work is found addressing the spectrum allocation/sharing problem for CRIoT using MEC technology except few researches which are based on cloud/fog computing in the CR context.

In [18], the proposed method utilizes the cloud to manage and allocate the free space of the licensed users to the unlicensed user by monitoring the spectrum utilizing the cooperate spectrum sensing and the sparse Bayesian algorithm to map the free space and convey it to the cloud data base and also uses a spectrum lessor to take care of the allocation of the free spaces to the unlicensed user. Then main disadvantages of this method are, first the use of cloud computing which has long latency since the powerful severs are always far away from the wireless devices. Second, the performance of the proposed protocol is not compared with any other protocol.

In [12], the authors propose a manifold learning dynamic spectrum allocation framework combining fog computing and cloud computing so that the received signal is processed close to where it is generated. Machine learning approaches are also incorporated to determine the available spectrum. Depending on the computational capabilities of the fog nodes, different features and machine learning techniques are chosen to optimize spectrum allocation. Each fog node has the ability to reason and choose the best spectrum candidate to transmit the signal without interfering with the licensed legitimate primary users. Also, the computing nodes send the periodic signal summary which is much smaller than the original signal to the cloud so that the overall system spectrum source allocation strategies are dynamically updated. Applying fog computing, the system is more adaptive to the local environment and robust to spectrum changes. The downside is that there is no communication between the Fogs Computing, which allows the appearance of certain conflicts between the decisions of the Fogs Computing which will be resolved by the Cloud that makes this solution less efficient. In addition, some decisions are made at the cloud level which generates a very high latency and therefore the performance of the solution will be degraded. This strategy was not implemented so no results are provided to approve its efficiency.

In [6], the study addresses the problem of Cognitive Radio Internet of Things (CRIoT) spectrum scheduling. The architecture of the used network is composed of three levels, IoT devices, fog nodes and a central fog server. A centralized time slotted algorithm was proposed, it uses discrete permutation particle swarm optimization for scheduling packets among predicted primary network empty time slots. An objective function is formulated to maximize the fairness index among fog nodes, minimize the packets’ queuing delay and minimize the number of dropped packets at each time interval. In order to evaluate the proposed protocol, it is implemented along with another protocol named spectrum auction. Using the fairness index, the average queuing delay and the number of dropped packets as evaluation metrics, simulation results proved that the proposed protocol outperforms the spectrum auction protocol. In the proposed scheduling algorithm, they use priorities to schedule SU requests. The authors do not indicate whether request priorities change during execution. If the priorities of the requests do not change, we can fall into starvation situations where some SU requests will not be processed. Another disadvantage of this approach is the use of a centralized scheduling technique which requires more time in downloading and uploading information over central fog server and if the latter breaks down this can lead to an interruption of all the process which demands a distributed scheduling algorithm to handle such huge data received from IoT devices. The authors did not evaluate their strategy in terms of communication cost.

In this article, the authors try to alleviate the spectrum allocation problem in the CRIoT network by assigning this task to more powerful nodes like MECs which can at the same time ensure the proximity of spectrum allocation services for the CRIoT nodes which pushed the authors towards MEC-based spectrum management architecture. Therefore, the proposed approach takes into consideration the economic aspect which has been neglected by most of the existing works of literature.

3. System model

3.1 Network topology

In this paper, the distributed architecture which applies MEC technology in cognitive radio network is proposed. This proposal addresses the spectrum allocation challenges to guarantee QoS requirements of CRIoT users and maximize the utilization of spectrum resources. As shown in Fig. 1, the CR network is composed of multiple cells with many CRIoT users who attempt to establish communication with the cellular Base Station (BS) in each cell. Therefore, the proposed architecture consists of the following tiers:

Figure 1.

MEC-aided CRIoT network for a distributed spectrum management.

IoT devices: These are smart mobile nodes within the CR network; there are two types of users in this system, primary and secondary users. Primary users have a licence for their spectrum usage while secondary users are the unlicensed users that can use both the unused licensed and unlicensed spectrum.

MEC server (Base station): Besides being cellular station for voice calls and data, it operates OpenFlow and can deliver MEC services. This MEC device acts as the local spectrum management unit that has the local spectrum information storage and plays an indispensable role in network management through the exchange of spectrum information with neighboring MECs.

Core Cloud: This layer represents cloud computing equipment which consist of several servers and data storage, which performs powerful tasks such as big data processing, data warehousing and service management.

3.2 Network assumptions and objectives

In this context, the authors assume that many BS are set up in the network, each of which covers one cell of the CRIoT users. Grouping nodes that share locally available spectrum holes can enable nodes to efficiently coordinate their local interactions and improves the efficiency of network functions such as spectrum management and balance energy consumption of CR users. In each cell, the MEC server is in charge of handling all the allocation requests and sharing proposals received from the SUs and PUs respectively to make a spectrum access decision and maximise a common objective function, taking into account the constraints of all users.

To overcome the problem of interference caused by opportunistic access to the spectrum, the authors assume that the PUs can cooperate with the SUs for licensed spectrum sharing. Thus, PUs are ready to share their unused channels with SUs for a specific amount of money and they send their sharing proposals to the MEC server. The authors assume also that the PU can share his channels with different SUs simultaneously to maximize spectrum utilization which generates the maximization of its gain.

In the proposed model, the spectrum allocation mechanism is restricted to the MEC server and the SUs are exempt from this mechanism. Hence, the SUs send their spectrum allocation requests to the MEC server for processing with possibility of paying the allocated channels. Furthermore, MEC server promote the use of licensed bands in accordance with the cooperative aspect between PUs and SUs, when an SU does not have the required price to pay PU, it will use the unlicensed band detected by the MEC server to accomplish its task.

Spectrum sensing process and PU proposals sending are considered proactive and not reactive to the SU-allocation demand. The proactive behaviour can reduce the response time of SU requests (to avoid wasting time). On another side, the authors assume that neighbouring MEC devices communicate directly with each other to exchange periodically information on unused spectrum in their cell and ensure the mobility of CRIoT users.

The main performance objectives of the above model are summarized as follows:

•
Spectrum utilization efficiency: the main goal is to maximize the spectrum utilization to satisfy SUs demands for spectrum sharing taking into account the overall system utility.
•
Energy efficiency: the energy consumption is the amount of electric energy used by CRIoT users. These devices have limited energy which requires balanced energy consumption.
•
Low latency: Latency is the time it takes for message to pass from IoT device to MEC server or from MEC server to IoT device. Therefore, it is necessary to minimize the delay time of SU which is the time it takes for accessing the available spectrum.
•
Low communication cost: In cooperative systems, communication should be targeted towards improving overall system utility. The number of messages exchanged by CRIoT users in the network for successful spectrum usage determines their communication cost.

4. Mec services based cooperative spectrum management

In this section, the structure of the proposed system will be explained in detail as well as the functioning of its various components. The researchers propose to introduce spectrum management services in MEC hosts which aim at maximizing the efficiency of resource allocation as highlighted in Fig. 2. These services work together sequentially to achieve the different stages of a spectrum allocation and simultaneously in order to process allocation requests in a reduced time. The researchers have assumed that the IoT device has a cooperative behaviour, for this a new component of cooperation is added in this device (Fig. 2). In the following, the authors will focus on the role of each component in IoT device and MEC server.

Figure 2.

MEC based spectrum management system.

4.1 IoT device structure

It is assumed that the IoT device contains the following components:

Sensing: CRIoT concept will consume considerable electrical energy especially in spectrum sensing phase. In order to conserve devices energy as long as possible, the detection of empty spectrum portions is assumed be restricted to the MEC server. However, CRIoT device will need to detect signals from BSs or other devices which can direct it to the nearest BS in order to transmit its request to it.

Mobility: it deals with mobility specific functionalities such as: tracking device’s location and performing mobility procedure with devices and base station. If IoT device wants to move or if the current spectrum cannot meet application-specific requirements, then it can request end-to-end connectivity in the network using this component which will subsequently implement the end-to-end path.

Cooperation: it produces the sharing proposals for the PUs which want to share their spectrum with the SUs. The proposal should be studied to set the main sharing parameters according to the state of PU. The proposal is in the form such that.

$\displaystyle\textit{RequestPU}(\textit{PUID},s,tp)$ (1)

where PUID is the primary user’s communication agent identification, $s$ is the amount of spectrum that PU is willing to give to SU, $t$ is the proposed spectrum holding time and $p$ is the expected price to get after sharing its spectrum.

Likewise, this cooperation component takes care to fix the SU needs taking into account its financial capacity for better service. A request is in the form such that.

$\displaystyle\textit{RequestSU}(\textit{SUID},s,tp)$ (2)

where SUID is the secondary user identification and it is used to help SU-leader agent to reply to the corresponding SU, $s$ is the amount of spectrum needed by the SU, $t$ is the desired time limit (or holding time) for the spectrum utilisation, and $p$ is the price it is willing to pay to PU.

Communication: IoT device can communicate with other devices in the network by sending and receiving messages which requires at least a single channel for communication. In the case of spectrum allocation, communication component of IoT device is responsible for sending allocation requests or sharing proposals and receiving responses on spectrum sharing.

4.2 MEC spectrum management

Recently, the concept of MEC services is on the rise wherein multiple services are provided to users as a whole including infrastructure, platforms and software applications. In this context, the researchers propose a new paradigm of providing spectrum management as a service on CRIoT infrastructure. The proposed solution is based on the cooperation between the different MECs that exist in the network where each MEC contains the following services:

Communication service: This service is used for communication between the MEC server and the others devices (SUs, PUs and MECs) in CR network. When receiving new sharing proposals or information on unused spectrum in other cells, it immediately sends them to the spectrum database for use in allocation process. When it comes to allocation requests, it will be sent to the scheduling service to take their requests into consideration.

After decision making, it should respond to the SU by providing it with the details of the allocated channels and also the PU should be informed that its channels are being shared. If the desired channels are offered by other MECs, then the communication service should contact them to confirm that they are still available for allocation. It is also the communication service which takes care of exchanging unused spectrum information with neighbouring MECs.

Sensing service: Employs energy detection, matched filter detection, or feature detection to monitor spectrum and identify spectrum frequencies used by PU/SU. Once the detection spectrum is launched, it starts the detection of empty spectrum holes and it should update the spectrum database with this new information. In the proposed approach, this service detects only the unlicensed channels because the unused licensed channels are sending to MEC server as sharing proposals by PUs. The channels detection process is performed on a periodic and adaptive manner, this period changes depending on the number of demands, if the number of demands is large in cell then the period is short, else the period becomes longer.

Scheduling service: it will insert each new SU request (allocation demand) in a list named “allocation list”, it will treat all requests of the same cell according to a custom scheduling algorithm proposed in the previous work [9] which is based on the sharing time and the energy of SU (Algorithm 1). Thus, the SU requests scheduling phase sets the priority of each request with a rank value and generates a request list by sorting the SU requests according to their rank values. The most priority request will be destined for the service analysis and will therefore be deleted from the list. The following algorithms use different performance metrics which are given in Table 1.

Table 1
The metrics definition

Symbol	Definition
AT	Arrival Time of SU.
EST	Estimated Sharing Time of SU to accomplish its applications.
EEn	Estimated Energy of SU from the beginning.
ERS	Estimated Request Satisfaction: the satisfaction value is calculated relative to the resources available in terms of
	the needed spectrum portion, price and sharing time.
EWT	Estimated Waiting Time: is the time that a SU can wait in the queue to share the needed spectrum.

Algorithm 1: Scheduling process by

\textit{MEC}_{c}

Figure 3.

Analysis process in $\textit{MEC}_{c}$ .

Analysis service: It analyses the spectrum database to identify all the channels that can satisfy the current request and provides output to the decision service. This output results contains either the set of licensed channels that check the sharing conditions as shown in Eq. (3), or the set of unlicensed channels if all licensed bands are busy or the SU wants to share the free bands. Moreover, the analysis service check directly the set of unused spectrum received from MEC neighbours to perform mobility of SUs. In addition, even SUs which are in the neighbourhood of other cells can exploit their sharing proposals. In all cases, if no appropriate channel is found, the request will be returned to the scheduling service and will be processed later (Fig. 3).

$\displaystyle\begin{array}[]{c}\textit{Sharing}\\ \textit{conditions}\\ \end{array}\mathop{\Leftrightarrow}\limits^{\forall SU,\forall PU}\left\{% \begin{array}[]{c}\textit{Channel required by SU request}\leqslant\textit{% Channels proposed by PU}\\ \textit{and}\\ \textit{Channel's using time by SU}\leqslant\textit{channels sharing time % proposed by PU}\\ \textit{and}\\ \textit{SU price offered}\geqslant\textit{PU price required}\\ \end{array}\right.$ (3)

Decision service: it is responsible for making the decision in selecting the suitable spectrum to be used by SU from the received set of channels. It calculates the estimated request satisfaction for all this channels (the value is relative to the needed spectrum portion, price and sharing time). Therefore, the appropriate spectrum resource is which has the maximum satisfaction value (Algorithm 2). Once the sharing decision is made, the information of allocated channels must be stored in the spectrum database and all parties concerned will be informed as well as the MEC neighbors.

Algorithm 2: Decision making by

\textit{MEC}_{c}

Figure 4 shows how the above services react to the reception of a new request.

Figure 4.

Spectrum management services working.

5. Experiments and results

5.1 Simulation setup

In this section, various numerical results are presented to evaluate the working of the proposed approach, based on the following simulation setup. The simulations are performed under the assumption of cognitive radio network with multiple sets of primary and secondary users. Besides, parts of access points (APs) are equipped with MEC servers. Considering the capacity of a single machine, the maximum number of primary and secondary users is 250 in total. Moreover, the rates $\lambda$ p and $\lambda$ s denote the distributions of primary and secondary users respectively. Then in our approach, each cell contains n SUs and m PUs. Each cell contains a MEC server deploys over it for spectrum management purposes.

In the simulation, the authors consider for each next round of $\lambda$ p there is an addition of one spectrum bands and set the bandwidth of each spectrum portion to 3.75 MHz [5]. According to [13], the mean spectrum usage of PUs is set to 40% and each PU has random unused spectrum portions. Also, the authors follow the assumption that once agreed, PUs would not be able to withdraw their commitments and they should share their spectrum with the corresponding SUs for the agreed time period. The network parameters chosen for evaluating the algorithm and the methodology of the simulation are shown in Table 2. In this work, the iFogSim framework is extended [10] to more appropriate simulate a MEC spectrum management experiment (as shown in Fig. 5). Each simulation is performed for a total of 10 runs and the average values are taken to plot the graphs. Hence, all the simulations are conducted in iFogSim overall PC with 2.4 GHZ dual processor and 3 GB memory.

Table 2
Simulation parameters

Parameter	value
Maximum number of SUs (N ${}_{S}$ max)	150
Maximum number of PUs (N ${}_{p}$ max)	100
Maximum Spectrum bands	10
Number of unlicensed channel per band	10
Size of a Spectrum portion (B)	3.75 MHz
PUs’ mean spectrum usage	40%
Number of simulation runs for each scenario	10

Additionally, during a simulation, the proposed distributed solution based on the use of MECs is compared to the cooperative approach based coalition (CAC) [9] and the approach based cloud computing (ACC) [18]. The CAC is our previous work in which the solution is distributed and based on cooperative multiagent system where agents are deployed on both SUs and PUs; these agents are grouped into coalitions where the powerful node is selected to be a leader. The leader of secondary user’s coalition organises the allocation requests based on their priorities using scheduling heuristics while the leader of primary user’s coalition handles the proposals for spectrum sharing. The ACC is centralized because it utilizes the cloud to manage and allocate the free space of the licensed users to the unlicensed users by monitoring the spectrum utilizing the cooperate spectrum sensing and using a spectrum lessor to take care of the allocation of the free spaces to the unlicensed users.

Figure 5.

iFogSim interface.

5.2 Obtained results

5.2.1 Successful spectrum allocation

Figure 6 depicts the average number of successful spectrum allocation. Thus, the simulation of the proposed solution and the other approach are run with several sets of SUs as showing in the figure.

The CAC and ACC have slightly a fewer successful spectrum allocations than the proposed approach, where almost SUs are fully satisfied because the proposed approach promotes the unlicensed spectrum utilization to avoid wasting time to find the appropriate PU’s proposal if it is not available, in other hand each allocation demand not satisfied in the time it be stored in the cache until the required spectrum will be available which shows the performance of the proposed approach.

Figure 6.

Impact of SUs number on successful spectrum allocation.

5.2.2 Response time

The time is one of the most important factors to be considered in CR network, for that the simulation is run with several values of SU agents. Figure 7 plot the average overall response times of SUs for a total of 15 to 150 SUs. The observed results show that the average response time value increases quickly with increasing of SU number and the proposed approach has a small value of overall response time compared to CAC and ACC.

In the proposed approach, the MEC services are closer to the SUs than the Cloud/fog services (CCA approach) and each service performs a precise task in a reduced time thanks to the power of the MEC server compared to a simple mobile node which takes care of all the management as it is proposed in CAC approach. The authors notice that the proposed approach has the optimal value of response time which means that SUs can easily find the unused spectrum showing the performance of the proposed approach in term of allocation process.

Figure 7.

Response time of SUs requests.

5.2.3 Communication cost

Figure 8 plots the histogram of average communication cost. The increasing pattern in communication cost is directly relational to the number of SUs. When SUs are less in number, the message exchange between the users is not high. Similarly, when number of SUs increases, there is whole information to share causing communication cost to increase. The authors notice that the communication cost for the proposed approach is upper than CCA which explained by exchanged messages of spectrum state between MEC neighbours in the proposed approach unlike CCA where all spectrum data is located in the cloud database which minimizes the communication cost.

Figure 8.

Impact of SUs number on communication cost.

5.2.4 Energy consumption

Some nodes consume less energy to accomplish their tasks while others consume much more, which results in earlier death of those particular nodes leading to shorter network life time. So, it is important to balance the energy consumption among mobile nodes to avoid node failures. Figure 9 illustrates the distribution of energy consumption on all networks, the CAC and ACC exhibit larger variance values while the proposed approach exhibits smaller variance of CR nodes energy because these nodes are exempt from spectrum detection and the waiting time is very little which reduces energy consumption then the proposed approach performs better, achieving a more uniform distribution of energy consumption across the network.

Figure 9.

Distribution of energy consumption of SUs.

6. Conclusion

This paper addresses the problem of CRIoT spectrum management. MEC is the key technology that the authors propose to remedy this problem. The main contribution of this work is the use of the distributed approach in which the spectrum management services is integrated in the MEC server for achieving high system performance. The cooperative behavior of the IoT device has led to a better use of the spectrum by avoiding interferences between users. Comparing with the distributed approach ACC based on a leader agent and the centralized approach CCA based on a cloud computing, the experimental results shows that the proposed approach based on the MEC can extend the lifetime of CRIoT and achieve a more uniform distribution of energy consumption across the network. Furthermore, the proposed solution works effectively and can absorb spectrum allocation demands without having higher latency and with average communication cost. In the future work, the authors will focus much more on the mobility of CRIoTs and data routing to improve the overall performance efficiency of the cognitive radio network.

Footnotes

Author’s Bios

	Fatima Zohra Benidris is a lecturer in Computer Science at Abdelhamid Ibn Badis University (CSTL Lab.), Mostaganem, Algeria. His research interests are: Cognitive Radio, mobile environment, Internet of thing, routing, Telecommunication network.
	Said Limam is an Associate Professor in Computer Science at Dr. Tahar Moulay University of Saida, Algeria. His research interests are: distributed system, grid computing, cloud computing, fault tolerance, data replication and consistency.

References

Khan

A.A.

Rehmani

M.H.

and Rachedi

, Cognitive radio based Internet of Things: Applications, architectures, spectrum related functionalities, and future research directions, IEEE Wireless Communications 24(3) (2017), 17–25.

Liu

Wang

Zhou

and Lu

, Energy-efficient cooperation in mobile edge computing-enabled cognitive radio networks, IEEE Access 7 (2019), 45382–45394.

Moon

, Dynamic spectrum access for internet of things service in cognitive radio-enabled LPWANs, Sensors 17(12) (2017), 2818.

Zhang

and Chen

, Selective spectrum leasing in Internet of Things via Nash Bargaining Solutions, in: Proc. IEEE 2nd Int. Conf. Cloud Comput. Intell. Syst, Hangzhou, China, 30 Oct.–1 Nov. 2013, pp. 841–845.

Kim

and Hossain

, Joint rate and power allocation for cognitive radios in dynamic spectrum access environment, IEEE Transactions on Wireless Communications 7(12) (2008), 5517–5527.

Tarek

Benslimane

Darwish

and Kotb

A.M.

, A new strategy for packets scheduling in cognitive radio internet of things, Comput. Networks 178 (2020), 107292.

Lam

Sheng

Hua

and Wang

, Advances and emerging challenges in cognitive internet-of-things, IEEE Transactions on Industrial Informatics 16(8) (2019), 5489–5496.

Benidris

F.Z.

Benmammar

and Bendimerad

F.T.

, An efficient spectrum allocation mechanism for cognitive radio networks, WSEAS Transactions on Communications 13(1) (2014), 534–547.

Benidris

F.Z.

Benmammar

and Krief

, A novel approach to spectrum management in cognitive radio networks using agent coalition formation, International Journal of Wireless and Mobile Computing 9(1) (2015), 27–35.

10.

Gupta

Dastjerdi

A.V.

Ghosh

S.K.

and Buyya

, iFogSim: A toolkit for modeling and simulation of resource management techniques in Internet of Things, edge and fog computing environments, Software: Practice and Experience (SPE) 47(9) (2017), 1275–1296.

11.

Salameh

H.B.

Almasri

Benkhelifa

and Lloret

, Spectrum assignment in hardware-constrained cognitive radio IoT networks under varying channel quality conditions, IEEE Access 7 (2019), 42816–42825.

12.

Chen

Shen

Pham

and Blasch

, Machine learning based intelligent cognitive network using fog computing, In Sensors and Systems for Space Applications X. International Society for Optics and Photonics 10196 (2017), 101960G.

13.

McHenry

M.A.

, NSF spectrum occupancy measurements project summary. Shared spectrum company report (2005). http://www.sharedspectrum.com.

14.

Ozger

Cetinkaya

and Akan

O.B.

, Energy harvesting cognitive radio networking for IoT-enabled smart grid, Mobile Networks and Applications 23(4) (2018), 956–966.

15.

Han

Gao

and Lu

, An effective multi-objective optimization algorithm for spectrum allocations in the cognitive-radio-based internet of things, IEEE Access 6 (2018), 12858–12867.

16.

Shigueta

R.F.

Fonseca

Ziviani

Viana

A.C.

and Munaretto

, A Strategy for Opportunistic Cognitive Channel Allocation in Wireless Internet of Things. IFIP Wireless Days, Rio de Janeiro, Brazil, 12–14 Nov 2014. 2015, 1–3.

17.

Kim

, Inspection game based cooperative spectrum sensing and sharing scheme for cognitive radio IoT system, Computer Communications 105 (2017), 116–123.

18.

Bindhu

, Constraints mitigation in cognitive radio networks using cloud computing, Journal of trends in Computer Science and Smart technology (TCSST) 2(1) (2020), 1–14.

19.

Ayoub

Z.A.

and Khader