With the increase of congestible frequency spectrums and the rapid development of wireless sensor networks (WSNs) applications, the new technology—combing cogitative radio technology with WSNs—called CWSNs will bring broad prospects to the field of radio and sensor networks. CWSN devotes itself to the solution of spectrum sharing in forms of networks, such as IEEE 802.11 and Bluetooth. A new algorithm is proposed for allocating the idle spectrum to secondary users (SUs). This algorithm uses “the last diminisher” algorithm, which appears in fair division models. It can help solve complicated problems in a simple fashion. This article combines the trust value and the method of spectrum allocation. The concept of trust value in reputation management is introduced. All the factors are applied to solve the realistic problems of spectrum allocation. With the growing number of sensor nodes, the stable throughput is of vital importance.
With the global data congestion increasing by 50 percent per year, the wireless sensor networks (WSNs), one of the fastest growing wireless networks in recent years, are faced with network traffic as well. Tens of thousands of tiny sensor nodes are distributed in the wireless sensor networks to detect surrounding data. With the rapid development of the WSNs, this potential application scenario has gradually proceeded to reality, from industry to home to civilians as well as the military. New wireless communication techniques, such as ZigBee and 802.15.4, have already enabled the interoperability between products to ensure extendibility of the network and low consumption.
Most of the wireless sensor networks work in unauthorized frequency bands, such as ISM, which is a universal 2.4 G frequency band. This frequency band is shared by many popular technologies, such as Wi-Fi and Bluetooth. Therefore, with the increase of applications based on the WSNs, the frequency band has become increasingly congested. The sharing of unauthorized frequency bands has become the theme of extended research [
Mitola first proposed the concept of cognitive radio in a paper [
The growing demand for wireless communication has become a great challenge [
Dynamic spectrum access can be used in WSNs. Therefore, a new type of network emerges, which is the CWSN. Typical wireless sensor networks manage finite spectrum and transmit perceived data to adjacent nodes. A fundamental aim of the wireless sensor network is to transmit large perceived data in real time and synchronously on the premise that the fewest possible resources are used. For example, in a WSN of the medical environment, a lot of nodes are used to monitor various signals of a patient and transmit them in real time.
Identically, with the growing development of the cognitive radio technology, the study of the security of the technology in various forms of networks is deepened. The trust mechanism and trust management have an irreplaceable position in the study of the cognitive network’s security. In a distributed wireless network, as there is no large base station, the security of the network depends on the cooperation between users.
As a hot area of wireless system research, trust management has an irreplaceable position in network defense and security. A crucial factor of trust management is the reputation value, and it is requisite to have a further analysis as well as discussion of the reputation value and combine it with necessary technology [
This way, CWSN is a new concept with the following advantages. Higher transmission range. Fewer sensor nodes required to cover a specific area. Better use of the spectrum. Lower energy consumption. Better communication quality. Lower delays. Better data reliability.
This paper is organized as follows. Section
The cognitive wireless sensor network (CWSN) is a new network form of cognitive radio technology based on the WSN. AS shown in Figure
CWSN construction.
Cognitive radio can perceive the wireless communication environment and change the system’s working parameters in real time and self-adaptively by a learning decision algorithm. It has two objectives: to improve the utilization rate effectively, to improve the communication reliability effectively.
Cognitive radio spectrum allocation assigns spectrums to one or more designated subscribers according to the number of users accessing networks and service requirements. The main purpose of spectrum allocation is to choose and utilize idle spectrums effectively through an adaptive strategy. The dynamic spectrum allocation strategy can effectively increase the flexibility of wireless communication, avoid conflicts between the authorized user and unauthorized user, fairly share spectrum resources, and meet user’s changing requirements for different services.
At present spectrum allocation has the following major models: graph coloring model, game theory model, auction model, and interference temperature model [
The graph coloring model in the graph theory is more mature when applied to planning and disposing frequency in the cell in the era of cellular communication. The available spectrum of the cognitive user is influenced by the location, work status, and the coverage of the main user due to its random use of the spectrum by the main user, which changes in the idle time.
The graph coloring model was proposed by Zheng et al. based on the graph theory. In the study of spectrum allocation, the network topology mapping of cognitive users is abstracted as the graph
The spectrum allocation in the cognitive system can be modeled as the throughput of the game, the player being the cognitive user, game strategy being the choice of the channel, and the game effectiveness related to the quality of the channel. The mathematical description of distributing spectrums using gaming is
Spectrum allocation has a lot of similarities to the auction in economics. For example, the objects which are bidden have a certain value, which is able to bring improvement in performance to the occupants. And they both have many competitors as well. So this theory can be applied to spectrum allocation in cognitive radio. Today the auction bidding theory has been proved to be an effective way to solve the problem of spectrum allocation.
The auction model is usually used in the centralized network structure. The central controller acts as an auctioneer and the secondary users as the bidder. In every auction, bidders bid a price for spectrum resources to meet their own demand, and the auctioneer will determine a winner as ruled. Compared with other allocation methods, the advantages of the auction bidding mechanism are that it is naturally dispersed and requires less signaling exchange and computational overhead.
Interference temperature equivalent to noise temperature is used mainly as a measure of the interference power and its bandwidth. It is introduced by FCC to quantify and manage interference. In this model, the cognitive facilities working in licensed bands can adjust the transmitter power and spectrum by measuring the interference environment to avoid interference to the authorized user exceeding the specified temperature threshold.
The interference temperature model uses the interference temperature as the decision threshold that is, used as the maximum cognitive user interference signal tolerated by the authorized users to make a judgment. In the working frequency, as long as the cognitive user’s accumulated radio frequency power does not exceed the set value (the threshold value refers to the maximum interference value tolerated by some frequency in a certain band), the secondary user can use this band. Any secondary user using this band must ensure that the interference to the authorized user does not exceed the threshold value.
Fair allocation [
Fair allocation is a mathematical theory on the basis of idealization and practical issues. The realistic problem is the allocation of divisible resources or materials to players who are potential owners of goods. The core issue of fair allocation is that the allocation should be conducted by players, who know the value of goods through negotiation or with the aid of an intermediary rather than an intercessor.
The fair allocation theory provides a definite standard for the division of various kinds of fairness. It aims at providing a program to implement fair allocation or proving impossibility and studying the disagreements between the theory and real life [
The assumptions about the valuation of the goods or resources are as follows: each player has this own opinion about the value of each part of the goods or resources, the value of a player of any allocation is the sum of his valuations of each part. Often just requiring the valuations to be weakly additive is enough, in the basic theory the goods can be divided into parts with arbitrarily small values.
There have been many allocation strategies for fair allocation, most of which conflict with one another. But they can still be combined. The fairness allocation mentioned above here means that every participant can gain goods with the smallest possible difference in amount or proportion.
Theoretically, spectrum resources can be divided into arbitrarily small channels and allocated to those who want to transmit data. Because of this, we borrow the “the last diminisher” algorithm and propose reputation based fair allocation for cognitive radio networks. This allocation strategy is capable of making on-demand and fair allocation for an arbitrary number of secondary users.
This chapter makes a detailed analysis of the spectrum allocation algorithm proposed in this paper. In the cognitive wireless sensor network, we assume that the user’s reputation value has been obtained through the trust management mechanism. And in a process of our spectrum allocation, the reputation value is fixed and cannot be changed. Assuming that the needs of secondary users are not exactly the same and that the quality of channels is not exactly the same, either, it is rational for secondary users to choose the appropriate channel for data transmission. All the above ensures the effective utilization of idle channels and the reasonable and fair distribution of channels between secondary users.
The fairness this article proposes is not the easy division in geometry, but rather matching one’s requirement with the piece of goods.
For an arbitrary number of participants, a balanced fair allocation algorithm named “the last diminisher” was proposed by Banach and Knaster in 1944 as follows. The first person cuts a slice they value as a fair share. The second person examines the piece.
If they think the piece is less than a fair share, they then pass on the piece unchanged.
If they think the piece is worth more than a fair share, they trim off the excess and lay claim to the piece. The trimmings are added back into the to-be-divided pile. Each remaining person, in turn, can either pass or trim the piece. After the last person has made his decision, the last person to trim the slice receives it. If no one has modified the slice, then the person who cuts it received it. Whoever receives the piece leaves with his piece, and the process repeats with the remaining people. Continue until only 2 persons remain; they can divide what is left by the divider-chooser method.
The fairness of the algorithm is that none of the players can diminish the amount of goods to a value smaller than the average value according to his or her will. It is said that the result of reduction might belong to oneself. Identically, in the case of denying the benefit of someone else, the piece of goods cannot be greater than average, either. Intellectually, everyone would divide goods into average pieces in their own opinion.
In the process of polling, the method of allocating and competing for “goods” on the basis of needs is coincidently parallel with the idea of pursuing optimization in the allocation of spectrums. When allocating spectrums, different users who need to transmit data have different requirements, such as the requirement for transmission rate and band-width. Meanwhile, it is well known that as a process of multiple users towards multiple users, the allocation of spectrums should be done in a certain order. For this reason, our original intention is to try our best to guarantee that all the users who need to transmit data are able to get appropriate channels by allocating channels in a certain order.
The node which needs data transmission is the center. All the nodes within the hop range form a group and then exchange reputation values at that time. In the period of time (denoted as
It’s not our responsibility to consider the production of these trust values. In order to ensure the stable operations of CRNs which support mobile computing and ubiquitous computing, the establishment of trust is an open, heatedly discussed, and challenging issue. A large amount of research has been done both at home and abroad. The authors in [
The following three parameters are used to determine the user demand: the transmission rate
The following three parameters are used to determine the channel level: the channel capacity
The time function is defined in (
The channel capacity
The duration
As shown in Figure
Agent architecture.
Order the SUs according to the rules:
The whole flowchart of the algorithm is shown in Figure
General flowchart of the algorithm.
As is shown in Figure
Channel grading.
Suppose that the data length to be transmitted at one point is
The priority function is shown in (
If
This part is shown in detail in Figure
Channel allocation of SUs.
If
Choose
Here, we assume that the time when the channel is idle has been gained in the active detecting method. For the reason that the throughput in the sensor networks is huge but the message included is about the external environment, the amount of data is small, and when taking security, efficiency, and confidentiality into consideration, different data types have different needs. Consequently, when choosing an appropriate channel, users have to choose those whose idle time is enough to do a complete data transmission.
Under condition 1, we choose the channels with the characteristic
Under conditions 2 and 3, we choose the transmission power
Set
The flowchart is shown in Figure
Channel Competition in SUs.
The second user with trust value
If
Set the competition value in (
Set
The competition is ended. All the second users that get channels can transmit messages in the channel, and other second users stay silent.
Every agent of the nodes gathers the feedback information after a transmission. Whether to increase or decrease the trust value is decided by feedback information which has reached the threshold or not.
If the SUs who have been allocated an idle channel correspond and transmit data normally, increase the trust value by fifty percent. If the SUs have the following behaviors, decrease the trust value to a half: having nothing to do with the allocated channel, always in the competition, destroying the communication function of the network in the form of replay attack, Dos attack, Sybil attack and so forth, leaking users’ privacy.
The trust value is related to order value, which decides the SUs’ sequence in allocation. When an idle channel is allocated to an SU, the competing SUs cannot take it into account. That is to say, the trust value determines the ownership of the idle channels.
Identically, the trust value also plays an important role in the field of trust management, and some more in-depth research can be done in the near future.
“The last diminisher” algorithm proposed by Banach and Knaster can solve the “Cake Cutting” problem impartially, but it cannot provide a satisfying solution for every user. Presume there is a piece of cake with multiple flavors, chocolate, cream, strawberry, and so forth, people who share the cake have different taste for cakes, and different parts of the cake have different values to different people. Geometrically equal distribution and simple equal distribution cannot solve the “Cake Cutting” problem. Therefore, this article uses “the last diminisher” algorithm for reference but does not distribute frequency spectrums totally depending on the algorithm. This article presents a method by which users can get satisfying frequency spectrums according to their particular needs.
Here, the frequency spectrum is the cake we mentioned before. We know different parts of the cake have different flavors. In other words, different idle channels have different parameters. Furthermore, users not only care about whether they can get a channel but also whether the channel they get can satisfy their needs. Therefore, “the last diminisher” algorithm cannot meet our requirements.
Banach and Knaster provided a useful thinking: polling for advice and then getting the “cut cake”.
The frequency spectrum distribution method this article presents can ensure the following aspects: user users who have high demands can select high-level channels; users who have low demands can select low-level channels, users with low reputation levels cannot get channels selected by users with high reputation values, and even if they finally get that channel, that channel is a degraded channel.
If users with a low reputation value or demands select a channel that users with high reputation value or demands need, then we regard these users as malicious users and reduce their reputation values according to the situation in the next round’s reputation value calculation.
In order to further explain that our method has outstanding performance, we build a simulation environment. The number of secondary users and available channels are given, the sensor network has numerous and densely distributed ground nodes, and thus the number of users are much larger than that of channels. In order to guarantee randomness and fairness of the simulation environment, every user has an initial random reputation value and random transmission data with the data length, data type, and transmission speed of the last jump. If no data is to be transmitted, then the data length is 0. For available channels, the channel capacity, idle duration, and maximum bearable transmission power of the channel are random in a certain range. Nearly all simulation parameters are random, which is most like the real sensor network environment.
We set the number of users as 100 and that of available channels as 20, and other parameters are random in a certain range and then perform 100 experiments. As is shown in Figure
Throughout in 100 users, 10 channels, and 100 simulations.
When it comes to practical simulation, we need to allocate channels to users following our algorithm. Random parameters may lead to
On the other hand, the utilization rate is a public focus. We combine the utilization rate and network throughput and conduct persuasive comparison. The utilization rate is the ratio of the number of channels allocated to users after one allocation to the total number of channels.
We set the number of users as 100 and that of channels as 10, 20, 30, and 40, respectively, and the results are shown in Figure The pattern of the channel utilization rate is similar to that of throughput, which means the throughput value will be greater if the channel utilization rate is higher. There is no fixed relation between the channel utilization rate and the number of simulation. When the number of available channels is quite small but the total number of channels is big, all channels will be distributed to secondary users, which is the spectrum utilization rate shown in Figure
Throughput and utilization rate in 100 users 10, 20, 30, 40 and channels.
As is shown in Figure
Spectrum utilization with trust values in 20 simulations.
This article analyzes and deals with the distribution of idle channels of different parameters to secondary users with different needs, with reference to the “Final Cutter” method. We basically guarantee every user can get the suitable spectrum, and the performance of our method is basically stable for the whole network. The variation of channel parameters and user needs caused by external disturbance will not influence the whole network’s status.
This article proposes an algorithm for dynamic spectrum allocation in CWSNs, which integrates the idea of “the last diminisher” in the field of fairness allocation, the demand of SUs, and the channel characteristics.
At the beginning of the article, we give a brief introduction to the necessity of the CWSN spectrum allocation and put forward our main idea. In Section
The WSNs are facing increasingly congestible frequency spectrums and strait channels. In dealing with the inadequacy of spectrum resources, we still have a long way to go. Further researches need to be done on the cognitive WSNs from the aspects of trust management, spectrum sharing, power control, and so forth.
This work is supported by the National Natural Science Foundation of China under Grant no. 61172068, the Program for New Century Excellent Talents in University (Grant no. NCET-11-0691), and the Fundamental Research Funds for the Central Universities (Grant no. K50511010003).