Abstract
Spectrum decision is the capability of Secondary Users to choose the best accessible spectrum band to satisfy a user’s Quality of Service (QoS) requirements. Spectrum decision comprises three primary functions; spectrum characterization, spectrum selection and dynamic reconfiguration of cognitive radio. The study of dynamic reconfiguration of transceiver parameters in spectrum decision making has been motivated because of its importance to the realization of efficient spectrum utilization and management in distributed mobile cognitive radio networks. Spectrum decision making in a distributed cognitive radio network is crucial, so as to ensure that an appropriate frequency and channel bandwidth are selected to meet the QoS requirements of different types of applications and to maintain the spectrum quality. In attempting to address the issue of dynamic reconfiguration of transceiver parameters in decision making for cognitive radio networks, different approaches can be found in the literature. However, due to some of the challenges associated with these approaches such as high computational complexity, ambiguity, non-repeatability and non-deplorability of these classical approaches, researchers are still trying to explore other techniques that will be less ambiguous, more efficient, more understandable and easier to deploy in a highly dynamic environment like distributed cognitive radio networks. Hence, this paper reviews the existing approaches, identifies the challenges and proposes a biologically inspired optimal foraging approach to address the decision making problem and other problems relating to the existing approaches.
Introduction
The recent explosion in multimedia, social networking applications and other technological innovations has witnessed the rising demands for broadband Internet connectivity. Rural and peri-urban users often encounter uneven transition with technology as they lack the proper infrastructure and technologies to access high-speed broadband connectivity.
In the recent past, multi-hop wireless technologies such as Wireless Mesh Networks (WMNs) have been proposed and utilized as the last-mile technology to extend broadband connectivity. Such technologies have also extended the connectivity of other wireless communication services to rural and remote areas where other means of connectivity are deemed prohibitive.
WMNs and other multi-hop wireless technologies often operate on dedicated radio frequency (RF) spectrum bands, mostly the unlicensed 2–5 GHz Industrial, Scientific and Medical (ISM) bands. Recently, the research has shown that utilization and saturation of these unlicensed spectrum bands are high in dense urban areas, thereby resulting in highly congested and unusable networks [1–3]. Hence, these technologies that operate in unlicensed spectrum bands are subject to high levels of radio interference and performance degradation. This is, especially when they are deployed in the urban areas and operating as ultra-dense networks (UDNs) in the 5 G radio access technology [4].
This exhaustion of usable RF spectrum bands has stimulated novel approaches in dynamic spectrum access and management techniques [4]. Recent advances in wireless networking technologies, such as cognitive radio (CR) technology, promise to address the issues (e.g. interference, network performance degradation etc) and challenges of the depletion and inefficient utilization of spectrum by optimally accessing the usable spectrum opportunistically and dynamically.
Cognitive radio networks (CRNs) have the capability to use the available or idle licensed spectrum intermittently or temporarily for communication in an opportunistic manner. Thus, a cognitive radio network (CRN) is an intelligent wireless transmission system which has the capability to change its transceiver parameters such as frequency and channel bandwidth based on its interaction (such as spectrum sensing) with the outside surroundings where it operates [2, 5]. This technology has a cooperative dynamic spectrum access and is capable of sharing spectrum amongst licensed/Primary Users (PUs) and unlicensed/Secondary Users (SUs). The CRNs operate under a bounded constraint that the PU transmissions should not be interfered with [6]. Hence, as soon as PU activities are detected on a given channel the SU must immediately vacate the channel and continue its transmission on another available channel [6].
In order to realise an efficient utilisation of spectrum in a CRN environment, a dynamic framework for spectrum management is required. This dynamic spectrum management comprises spectrum sensing, decision making, sharing and spectrum mobility, as depicted in Fig. 1.
Spectrum sensing includes identifying spectrum holes and the capability to quickly detect the arrival of PU transmissions in the spectrum hole possessed by the SUs. Spectrum sharing alludes to coordinated access for the channel selected by the SUs while spectrum mobility is the ability of a cognitive radio to leave its currently occupied channel when it detects PU. Spectrum decision is the ability of the SUs to choose the best accessible spectrum band to fulfil users’ Quality of Service (QoS) requirements. Spectrum decision comprises three major functions: spectrum characterization, spectrum selection and dynamic reconfiguration of cognitive radio [7].
Similar to traditional wireless networks, a CRN topology can be classified as either centralized (infrastructure-based) or distributed (infrastructure-less or ad-hoc based) network topology, as depicted in Figs. 2 and 3 respectively. In the centralized environment a central node such as a base station or access point is deployed with several SUs associated with it. In contrast, the distributed topology requires that the SUs communicate directly with each other without any central or controlling node.
Unlike the centralized and traditional wireless ad-hoc networks, where the number of supported channels is fixed and low (mostly less than ten or at most in order of tens). The spectrum decision in a distributed CRN, where the supported numbers of channels ranges in the order of thousands is a serious challenge. This challenge needs to be addressed appropriately [3, 9].
In a distributed CRN environment such as in a mobile ad hoc network, when there are several frequencies and channel bandwidth available, how to dynamically choose the best one for the secondary user based on the spectrum quality and the QoS requirements of various types of applications is an important challenge.
A low frequency signal (e.g., 900 MHz) can travel a greater distance and penetrate walls and other obstacles, but its information transmission capacity is lower and the precision in deciding direction of signal arrival is poorer [9]. However, a higher frequency signal (e.g., 5 GHz) can only travel a shorter distance, but has the capability to convey more information and to exhibit directionality. The spectrum bands’ diversity and the guiding principles by the communications regulatory agencies on how to access the spectrum implies that the CR nodes for mobile ad-hoc networks should dynamically adapt their operating frequency and channel bandwidth techniques.
In a CRN, the SU nodes may find the frequency and channel bandwidth availability to be high at a particular location at some point in time but very low at another time in the same location, due to the PU’s activity. This variability in frequency and channel availability contradicts traditional ad hoc networks, where the network operates on a pre-decided set of frequency and channels, which remain unchanged over time. Hence, the ability of the CR nodes to decide on which frequency and channel to communicate and to dynamically reconfigure themselves, becomes an important issue that needs to be addressed in order to achieve the spectrum-efficient utilization goal of the CRN [4].
The dynamic reconfiguration of both the operating frequency and channel bandwidth in a distributed CRN has not received sufficient attention despite their importance in spectrum decision making [3, 9].
The existing literature [10–13] has focused on spectrum decision making for CRNs. However, to the best knowledge of the authors, there is no previous studies in the literature dealing with the analysis of reconfiguration of spectrum parameters for distributed CRNs. Thus, this current paper envisages to contribute in the following areas: Providing the need for studying distributed reconfiguration of the spectrum parameters for decision making in CRNs; Detailing the different approaches that other researchers have applied previously to address the problem of dynamic reconfiguration. Proposing a biologically inspired foraging approach to address the problem of dynamic reconfiguration in mobile distributed CRNs.
The remainder of this paper is organized as follows: Section 2 presents the Spectrum decision making and the Centralized vs. Distributed Spectrum decision making in CRNs, while Section 3 presents the existing approaches and their challenges, in solving the problem of dynamic reconfiguration in both centralized and distributed CRNs. Section 4 discusses the proposed model, using a biological-inspired optimal foraging approach. The paper is concluded in Section 5.
Spectrum decision making in CRNs
The spectrum decision is the capability of a CR to choose the most appropriate available spectrum band to satisfy the secondary user’s communication and quality of service requirements, without causing any harmful interference to the primary users. The spectrum decision in CRNs comprises three main stages namely: spectrum characterization, spectrum selection and parameters reconfiguration.
In CRNs, where there is a wide frequency range, there exist many spectrum bands with different channel characteristics. However, in order to select the most appropriate spectrum band it is very important to identify the characteristics of each spectrum band that are available. Hence, spectrum characterisation allows the SUs to characterise the spectrum bands by considering the strength of the signals received, level of interference and the number of users currently residing in the spectrum [2].
After the identification and characterization of spectrum holes, the next step in CRN decision making is the selection of the best available spectrum that is suitable for the SU’s communication and specific QoS requirements. In mobile CRNs the set of available channels for each user is not static. This phenomenon makes both the network topologies and radio frequency propagation keep changing as the mobile CR station changes its positions. For this reason the spectrum selection approaches in cognitive radio for mobile ad hoc networks should be closely coupled with routing protocols. A detailed analysis of spectrum characterization and spectrum selection can be found in [2].
The dynamic reconfiguration of transmission parameters occurs after the spectrum has been characterized and selected. Cognitive radio reconfiguration techniques enable the secondary users to dynamically reconfigure its transmission parameters (such as operating frequency, channel bandwidth and transmission power) for optimal operation in a certain spectrum band, so as to satisfy the user’s QoS requirements. The SU’s QoS requirements includes application requirements, such as throughput and latency, operational policies and environmental conditions requirements. Potentially, a number of approaches can be used to determine how the radio frequency and channel bandwidth parameters settings affect the network performance.
Most of the existing research in the area of dynamic reconfiguration of transmission parameters (operating frequency and channel bandwidth) has focused on centralized-based CR [9]. This can be attributed to the fact that it is easier to set-up and to manage a centralized-based CRN since almost all the computational work is done on a central control system or base station. Also, the centralized CRN operational cost is minimal compared to that of distributed CRNs. However, some of the issues with a centralized system include failure or link disconnection on the part of central system, which will automatically mean that the entire network cannot function effectively. Another related issue is the applicability of a centralized control system in a mobile ad hoc network environment, where nodes can leave and join the network at any time.
The recent increase in the number of potential mobile ad hoc network applications such as, in a military battle-field communication and a natural disaster relief, the need for a reliable communication of nodes are inevitable. Such applications have compelled the current research attention to be directed toward investigating effects of dynamic reconfiguration of transmission parameters on spectrum-efficiency performance of the distributed CRNs.
The next section reviews the existing approaches and their challenges, which other researchers have used to address the issue of dynamic reconfiguration of transmission parameters both in centralized and distributed CRNs.
Existing spectrum reconfiguration approaches and their challenges
In an attempt to address the problem of dynamic reconfiguration of transmission parameters in CRNs different approaches have been proposed. These approaches can be categorized into 3 main groups, namely, theoretical, statistical and predictive, and biological approaches.
Theoretical approach
This approach uses theoretical analysis to evaluate network performance. The analytical hierarchy process and queueing theory are the most common theoretical approaches used in CRNs parameter reconfiguration. In [14, 15], an automatic spectrum characterization and operating frequency reconfiguration framework for distributed CRNs was proposed. Their proposed framework was based on the Analytic hierarchy process (AHP). The framework was evaluated using simulation and three performance metrics were measured; spectrum decision and reconfiguration, throughput and spectrum handoff rate for secondary users. AHP is a popular method, useful for making complex decisions. However, its high computational complexity and the uncertainty related to the judgements in the pairwise comparison-matrix, which are not considered, limits the application of the AHP result in a highly dynamic environment like cognitive radio mobile ad hoc network.
A minimum variance-based (MVSD) and maximum capacity-based spectrum decision making (MCSD) framework was proposed in [3]. The proposed framework was used to determine the spectrum bands usage, based on different applications’ requirements in a centralized cognitive radio network. Based on the results of their study it was observed that the proposed MCSD introduced some additional frequency-channel switching delay, which led to some degradation in the QoS. The provisioning of good QoS is of high importance in a mobile ad hoc network environment and having a relatively high frequency-channel switching delay will lead to poor network performance. The high switching delay experienced in their study can be attributed to the complexity in the mathematical constructs. This makes the consideration of such a framework in a real world environment very difficult.
It can be observed that the existing theoretical approach for spectrum parameter reconfiguration in CRNs has some degree of high computational complexity which limits their applications in a dynamic real world environment. In order to address this problem, researchers would need to reduce the number of steps involved in computation, so as to reduce the switching delay time and to improve the network throughput. Also, the researchers could look at combining the advantages of AHP with those of MVSD.
Statistical and predictive approach
Weingart et al. [16] proposed a parameter reconfiguration predictive model based on Design of Experiment (DOE), which is an asset of tools and methods for determining cause and effect relationships within a system, and the analysis of variance (ANOVA) statistical approach. The model was used to develop a reconfiguration method for the CRNs, based on the communication requirements. Some of the challenges that the authors tried to address include: (i) making decisions about how and when to change the configuration parameters of the network communication layers (application, data link, routing and physical layers); (ii) how the reconfiguration changes will be propagated throughout the entire network; and (iii) the amount of time spent in computing the new reconfiguration.
The model was tested using the OPNET simulator and four performance metrics were considered: latency, jitter, bit loss and throughput. The DOE method applied in their study uses periodic data gathered from the central system. Thus, the performance at every point in time is only based on the last data collected. One of the challenges with this kind of approach is that it is not dynamic, and since the performance is only based on the last periodic data it makes most of the reported result unreliable.
In a mobile ad hoc environment, where nodes join and leave at any time, making decisions based on last collected periodic data makes such results unreliable especially, during situations like war or natural disaster. It would be interesting if the work could be improved by making the model prediction based on real time data collection.
A predictive model based on Location Aware Spectrum Database (LASD) was proposed in [17]. The main goal of the authors was to develop efficient spectrum utilisation in TV white space. Limpopo province in South Africa was used as a case study. In the proposed model the usability characteristics of a TV channel for the purpose of opportunistic transmission was measured based on Limpopo TV white space statistics. In their model both data collection and decision making are done by the central system. The threshold mechanism was used by the LASD model to separate channels with frequent primary users’ emergence. The model proposed in their study is location based and, due to the dynamic and transient nature of distributed mobile ad-hoc networks, the data used to develop the model can change at any time. Hence, it makes the developed model difficult to implement in a different location. Also, for such result to be reliable, there will be a need for continuous updating of the database information.
The parameters reconfiguration in CRNs reviewed thus far used theoretical [18], statistical [17, 19] and common control channel [9] approaches. The review pointed out some of the disadvantages observed, such as high computational complexity in theoretical methods, repeatability, and the analysis approach in statistical and predictive methods. All these challenges limit the practical implementation and usability of the aforementioned approaches in a highly dynamic distributed environment like cognitive mobile ad hoc network. Hence, there is still the need for an optimal approach in solving the problem of dynamic reconfiguration of transmission parameters.
In order to address some of the problems with the existing traditional approaches, researchers are exploring the game of choosing [20], fuzzy logic [21], Bayesian theory [22] and biological [23–26] approaches.
The game of choosing approach uses a psycho-sociology concept in order to make decisions and reconfigure the transmission parameter. One of the major problems with this approach is ambiguity in the number of processes involved in decision-making. This in-turn leads to high communication overhead and low throughput. The fuzzy logic and Bayesian approaches use some artificial intelligence concepts in addressing the challenge of transmission parameter reconfiguration in CRNs. However, the major challenge with these approaches is that, they are highly computationally intensive. In distributed cognitive mobile ad hoc network, the SUs are only opportunistic users. Therefore, the transceiver parameters reconfiguration in such an environment should be executed without unnecessary delay. In order to achieve a low communication delay overhead in parameters switching and reconfiguration, the approach to be used should be analytically simple and applicable in generic situations.
The biologically inspired approach has been described and been adopted by many researchers in the field of communication networks, due to its analytical simplicity and its generic applications. Different biologically-inspired approaches (swarm optimization, anti-colony optimization, collective robotic systems, and optimal foraging theory) have been proposed in other areas of wireless networking. However, based on the results from previous studies [24, 28] in other wireless networking in terms of applicability, simplicity and generic properties, we shall adopt the use of a biologically-inspired optimal foraging approach in addressing the problem of parameter reconfiguration in a distributed cognitive mobile ad hoc networks.
Proposed approach: Optimal foraging approach
The optimal foraging theory is the study of how natural foraging organisms in a random environment make optimal decisions that would help them to maximise efficiency, such as supplements utilization, survival likelihood, long lifetime and minimizing possible dangers.
Foraging theory uses different models to characterize how solitary foraging animals search for prey types and make optimal decisions on which prey to eat or not, in order to maximise their efficiency [29]. The characterizations of nutrients intake by foraging animals was adopted and modeled as an optimization process, which is now generally referred to as optimal foraging theory. One of the main advantages of the optimal foraging concept is that it allows a forager to make optimal decisions on the most appropriate prey types to eat, in order to maximise their efficiency within the minimum time interval. Optimal foraging theory can be applied in the area of decision making for CRNs to model the SU’s operating frequency and channel bandwidth reconfiguration as depicted in Fig. 4.
There are many existing biologically-inspired foraging models [30–33] but the optimal composite model of foraging theory is recommended because of its effectiveness and applicability to decision making in terms of parameters reconfiguration for the distributed cognitive mobile ad hoc network.
One of the key factors that determine foraging efficiency is the criterion that the forager uses. Foragers should seek to match their search efforts to the anticipated relative profitability’s earned from different parts of their environment. In a distributed cognitive mobile ad hoc network, where nodes can leave and join the network at any time, using the random search for prey type by dividing the search into intensive and extensive search modes will produce an efficient result within the minimum time.
In intensive search mode, the SU’s node will search an area thoroughly by taking short step lengths with frequent reorientations. This search mode can be employed in resource-rich areas such as rural areas, where there are many available frequencies, which the SU’s can use opportunistically.
However, for the extensive search mode, the SUs node moves efficiently across resource-poor areas by making long steps with few interruptions. This search mode can be employed in resource poor areas such as urban or suburb areas, where there are very few available frequencies, which can be opportunistically used by the SUs.
The combination of both intensive and extensive search modes is known as the optimal composite search model. Our model will utilise this combination of modes to search and to decide on the frequency to use for communication. In our model, the SUs will start searching for available frequencies using intensive mode, and after a specified period, if no frequency is available, the SUs will switch to extensive searching mode. This hybrid searching will help to reduce the switching delay, maximize throughput and make the reconfiguration of spectrum parameters faster, since it doesn’t involve many criteria or complex computation in selecting a frequency.
Conclusion
The study of dynamic reconfiguration of transceiver parameters in spectrum decision making has been motivated because of its importance to the realization of efficient spectrum utilization and management in distributed mobile CRNs. Spectrum decision making in a distributed cognitive radio network is crucial in ensuring that an appropriate frequency and channel bandwidth are selected to satisfy the QoS requirements of different applications and to maintain the spectrum quality.
In attempting to address the problem of dynamic reconfiguration of transceiver parameters in decision- making for CRNs different approaches (predictive, theoretical, statistical, CCC, etc.) have been applied by the researchers. However, due to challenges like high computational complexity, ambiguity, repeatability and applicability of these classical approaches, researchers are still trying to explore other approaches that will be less ambiguous, more efficient, understandable and easy to deploy in a highly dynamic environment like distributed cognitive mobile ad hoc networks. The biologically-inspired approach such as foraging theory, has been described and is being adopted by many researchers in the field of communication networks, due to its advantages. These advantages include analytical simplicity, optimum solution, flexibility, adaptation, robustness (ability to communicate, even when one or more nodes fail) and its generic applications. However, this approach has not been explored or tested in dynamic reconfiguration of spectrum parameters, for either centralized or distributed CRNs topology. Hence, this paper proposes a concept of biologically-inspired foraging approach to address the decision making problem and other problems relating to the existing approaches. In addition, optimal foraging will contributes to the achievement of an efficient spectrum- aware cognitive mobile ad hoc networks. In future, we plan to implement our proposed model and test it using computer simulations.