GB2465756A - Spectrum allocation in cognitive radio networks - Google Patents

Abstract

Medium access coordination and spectrum allocation for establishment of wireless communications in the context of dynamic spectrum access for cognitive radio networks. For a given network topology, the proposed method first generates an interference graph (or conflict graph) 24 based on the network topology, traffic flows and interference constraints. Then a graph colouring algorithm 26 is applied to the interference graph and the chromatic number of the graph is obtained. The chromatic number is the minimum number of distinct spectrum slices required. The spectrum allocation is also based on a max-min fair share algorithm 32 or proportional fair algorithm 30 according to the spectrum demands for different spectrum slices or blocks. The proposed technique helps achieve two goals in cognitive radio networks: efficient utilization of spectrum opportunities and fair allocation of spectrum resources among secondary users.

Description

The invention relates generally to medium access for establishment of wireless communications in a network. More particularly, but not exclusively, it relates to aspects of the technology in the context of spectrum allocation in cognitive radio networks.

The proliferation of versatile wireless services has increased demand for radio spectrum, a scarce resource much of which is licensed for specific uses or to specific groups of users. However, studies have shown that there exist spectrum holes where a band of frequencies assigned to a primary user is not being utilized by that user at a particular time and specific geographic location. (The term "primary user" generally refers to the licensed user of the spectrum or a user recognised as having high priority for the spectrum band.) The implication of this observation is that efficient spectrum usage can effectively be viewed as a spectrum access problem.

Recently, Dynamic Spectrum Access (DSA), or opportunistic spectrum access, has emerged as a way for secondary users to exploit spectrum holes (also known as "white spaces") to significantly improve spectrum utilization. A secondary user is a user who is authorised to use licensed spectrum opportunistically, without causing unacceptable interference to primary users. Generally, a secondary user accesses spectrum on a temporary basis when primary users are not making use of the spectrum, and should exit once a primary user arrives. In this context, there is an increasing interest in applying the concept of cognitive radio to DSA networks.

Cognitive radio is a field of wireless communications technology in which either a network on a distributed basis or a wireless node in particular can change parameters governing transmission or reception characteristics, such as operating frequency, modulation waveforms, and transmission power. The alteration of parameters can be based on active monitoring of several factors in the external and internal radio environment, such as reservations made of the radio frequency spectrum, user behaviour and network state. In addition, cognitive radio networks and devices are typically implemented such that effective communication is established without creating undue interference to other networks.

Conceptually, the idea behind DSA is simple -when a secondary device identifies spectrum holes (opportunities) in the spectrum it wishes to use, it can transmit using those opportunities in a manner that limits the interference perceived by primary users.

However, the realization of DSA currently faces several challenges. These include wide-band sensing, opportunity characterization and identification, efficient spectrum allocation, coordinating protocols for multiple nodes, and definition and application of interference-limiting policies.

Known approaches to regulating spectrum allocation can be classified broadly into two strategies: centralized and distributed schemes.

Centralized methods for managing and coordinating spectrum access, in which a central controller allocates spectrum to all nodes of a network, have been put forward by the IEEE 802.22 working group on Wireless Regional Area Networks ("WRAN5") (www.ieee802.org/22I), and in papers such as: "DSAP: a protocol for coordinated spectrum access," (V. Brik, E. Rozner, S. Banerjee, and P. BahI, First IEEE International Symposium on New Frontiers in Dynamic Spectrum Access Networks, DySPAN 2005, 8-1 1 Nov. 2005, pp. 611-614), and "DJMSUMnet: New Directions in Wireless Networking Using Coordinated Dynamic Spectrum Access" (M. M. Buddhikot, P. Kolodzy, S. Miller, K. Ryan and J. Evans, Proceedings of the Sixth IEEE International Symposium on World of Wireless Mobile and Multimedia Networks, 2005, pp. 78-85). The whole contents of these two documents is incorporated herein by way of reference.

In particular, in 802.22 a cognitive radio network consists of multiple cells. Within each cell, there is a base station (BS) that supports a set of fixed wireless subscribers called customer premise equipments (CPEs). The spectrum of interest is divided into a set of non-overlapping channels which a BS can use to serve its corresponding CPEs. The spectrum is actually licensed to a set of primary users (PUs). More detail is described in a paper "Maximizing Spectrum utilization of Cognitive Radio Networks Using Channel Allocation and Power Control" by A Hoang and Y Liang, IEEE VTC Fall, 2006, the whole contents of which are incorporated herein by way of reference. The 802.22 MAC has many similar features as those of 802.11 and 802.16. However, it also defines several MAC operations specifically tailored for dynamic spectrum access, such as initial connection establishment and. incumbent detection. Details can be found in IEEE 802.22 WRAN WG referred to above.

For distributed schemes, the literature contains several papers on MAC and network routing protocols for cognitive radio networks and multi-channel networks.

In "Collaboration and Fairness in Opportunistic Spectrum Access," (H. Zheng, C. Peng, IEEE International Conference on Communications (ICC), May 2005, pp. 3132-31 36), available spectrum channel is mapped to a colour, with a set of vertices denoting the users that share the spectrum. The problem then reduces down to colouring each vertex (allocating channels) while trying to maximise a "utility function". The whole contents of this document is incorporated herein by way of reference.

In "MAC-layer Scheduling in Cognitive Radio Based Multi-Hop Wireless Networks," (M.

Thoppian, S. Venkatesan, R. Prakash, R. Chandrasekaran, International Workshop on Wireless Mobile Multimedia, Proceedings of the 2006 International Symposium on World of Wireless, Mobile and Multimedia Networks, 2006, pp. 191-201), the optimal MAC-layer schedule problem is formulated as an Integer-Linear Programming (ILP) problem, which is known to be computationally complex. Further, the disclosed approach does not accommodate mobility and dynamic node arrivals and departures since it assumes a static multi-channel wireless network. The whole contents of this document is incorporated herein by way of reference.

The authors of the paper "Collaboration and Fairness in Opportunistic Spectrum Access" referred to above propose a label-based progressive minimum neighbour first (PMNF) graph colouring approach for utility-based channel allocation in cognitive radio networks. The paper "MAC-layer Scheduling in Cognitive Radio Based Multi-Hop Wireless Networks' describes an optimization framework for channel assignment. A paper entitled "Frequency allocation for WLANs using graph colouring techniques", IEEE WONS, 2005 by J. Riihijaryi et al, whose contents incorporated herein by reference, uses graph colouring for IEEE 802.11 AP channel allocation.

According to an aspect of the present invention there is provided a method of determining spectrum allocation in a wireless network comprising the step of determining the spectrum allocation required for the wireless network by means of a graph colouring algorithm.

The wireless network is typically a cognitive radio network whereby the step of determining the spectrum allocation involves determining a number of different spectrum slices or blocks required for the cognitive radio network based on the graph colouring algorithm.

In one embodiment the method comprises generating an interference graph based on at least one network parameter and applying the graph colouring algorithm to said interference graph to obtain the chromatic number of the graph. The network parameter may be selected from a group including network topology, traffic flows and interference constraints.

In a further embodiment the spectrum allocation is based on a max-mm fair share algorithm or on a proportional fair share algorithm.

In an embodiment the graph colouring algorithm comprises the steps of:- (a) initializing the degrees of saturation (DS) of all vertices to zero; (b) selecting an uncoloured vertex of highest DS; (C) colouring the selected vertex in a greedy manner using the smallest colour admissable; (d) updating the DS of the uncoloured vertices neighbouring the one coloured in the step (c) and (e) if all vertices are not coloured returning to step (b) The graph colouring algorithm is for example a DSATUR algorithm In a further aspect of the invention there is provided a wireless network operable to establish wireless communication between a plurality of wireless communication apparatus in a wireless communications medium defining a plurality of channels, the network comprising channel availability determining means operable to determine whether any of said one or more channels are available for use by one or more of the wireless communication apparatus; channel selection means operable to select for said one or more wireless communication apparatus, at least one channel available for use based on an available capacity of said at least one channel; and wherein the channel availability determining means is operable to determine said at least one channel allocation by means of a graph colouring algorithm.

Typically the wireless network is a cognitive radio network, and the channel availability determining means is operable to determine the number of different spectrum slices or blocks required for the cognitive radio network based on the graph colouring algorithm.

The channel availability determining means may be operable to generate an interference graph based on at least one network parameter and applying the graph colouring algorithm to said interference graph to obtain the chromatic number of the graph. The parameters selected from the group including network topology, traffic flows and interference constraints.

The wireless network can have the spectrum allocation based on a max-mm fair share algorithm or on a proportional fair share algorithm.

In one embodiment of the wireless network the graph colouring algorithm, for example a DSATUR algorithm, comprises the steps of:- (f) initializing the degrees of saturation (DS) of all vertices to zero; (g) selecting an uncoloured vertex of highest DS; (h) colouring the selected vertex in a greedy manner using the smallest colour admissable; (i) updating the DS of the uncoloured vertices neighbouring the one coloured in the step (C) and (j) if all vertices are not coloured returning to step (b) In a further aspect of the invention there is provided a storage medium storing computer executable instructions which, when executed on a general purpose computer controlled wireless network, causes the network to become configured to perform the method of the invention defined above.

There is also provided a signal carrying computer receivable information, the information defining computer executable instructions which, when executed on a general purpose computer controlled wireless network, causes the network to become configured to perform the method of the invention defined above.

In a yet further aspect of the invention there is provided a storage medium storing computer executable instructions which, when executed on a general purpose computer controlled wireless network, causes the network to become configured to a wireless network of the invention as defined above.

In a yet further aspect of the invention there is provided a signal carrying computer receivable information, the information defining computer executable instructions which, when executed on a general purpose computer controlled wireless network, causes the network to become configured to a wireless network of the invention as defined above..

Further preferred features of these aspects of the invention will now be set forth by way of the following description of specific embodiments of the invention, provided by way of example only, with reference to the accompanying drawings in which: Figure 1 is an example of the application of graph colouring using four colours; Figure 2 is a flow diagram of an algorithm for a graph colouring method based on DSATUR; Figure 3 is an example of an 802.22 cognitive radio network; Figure 4 is an example of a topology graph and its associated interference graph; Figure 5 is a flow diagram of an algorithm for spectrum allocation according to an embodiment of the invention and Figure 6 depicts an exemplary representation of a spectrum available to an opportunistic user.

Specific embodiments of the present invention will be described in further detail on the basis of the attached diagrams. It will be appreciated that this is by way of example only, and should not be viewed as presenting any limitation on the scope of protection sought.

Figure 1 is an example of the application of graph colouring using four colours. Graph colouring is a known technique for assigning labels traditionally called "colours" to elements of a graph subject to certain constraints. In its simplest form, it is a way of colouring the vertices of a graph such that no two adjacent vertices share the same colour. The colours represent the constraints for a particular application, for example frequency allocation in a communication network whereby to minimise interference between adjacent slices of the available spectrum, each slice of the spectrum is assigned a colour such that no two adjacent vertices share the same colour. The least number of colours needed to colour a graph is called its chromatic number. In the graph depicted in Figure 1 the chromatic number is 4.

In this description "graph colouring" will be referred to in the context of colouring of vertices of a graph. It will be understood by a person skilled in the art that other graph colouring techniques using the same basic principle can be adopted, for example edge colouring so that no two adjacent edges share the same colour or face colouring whereby a colour or region is assigned to each face or region of a planar graph so that no two faces that share the same boundary have the same colour.

In an embodiment of the present invention, having once identified a given network of secondary devices, the method includes constructing an interference or conflict graph G(V, E). In centralized networks such as IEEE 802.22 (as shown in Figure 3), the base stations (BS) are the vertices of the graph. If two BSs interfere with each other on the same frequency band (e.g. within a certain distance), then there is an edge between them. An objective of spectrum allocation is to minimize the number of distinct spectrum slices required while assigning different spectrum slices to any two vertices connected by an edge. In distributed networks, the vertex set of the interference graph is the set of traffic flows in the topology graph. If two flows interfere with each other on the same frequency band, then there is an edge between the two vertices.

An example is shown in Figure 4a in which the topology graph depicts five BSs, 1 to 5.

The traffic flows between BSs 1 and 2 is depicted by line A; the traffic flow between BSs 1 and 3 is depicted by line B; the traffic flow between BSs 2 and 3 is depicted by line C, the traffic flow between BSs 3 and 4 is depicted by line 0 and the traffic flow between BSs 4 and 5 is depicted by line E. This topology graph is then transformed into an interference or conflict graph 4b in which each of the traffic flows A, B, C, D and E become vertices of the interference graph, and the lines between each pair of vertices illustrates there is interference between those respective pairs of traffic flows The above problem is a typical graph colouring problem, which is NP-hard. Specifically, a vertex colouring is an assignment of labels or colours to each vertex of a graph such that no edge connects two identically coloured vertices. The most common type of vertex colouring seeks to minimize the number of colours for a given graph. The minimum number of colours with which the vertices of a graph G may be coloured is called the chromatic number, denoted by X(G).

There exist a number of heuristics, one of which is DSATUR (degree of saturation).

This is described in a paper entitled "New Methods to colour the vertices of a graph, Corn. of the ACM, 1979 by D. Brelaz, the whole contents of which are incorporated herein by way of reference. It is a simple and yet effective graph colouring method. A flowchart of the algorithm is shown in Figure 2. Degree of saturation is defined as the number of colours used in the neighbourhood of a vertex. At the beginning of the algorithm, depicted as step ID in Figure 2, the degrees of saturation of all vertices are set to zero. An uncoloured vertex with the highest degree of saturation is chosen. In case there are more than one vertex with the same degree of saturation, the one with the highest number of uncoloured neighbours is selected, shown as step 12. The selected vertex is further on coloured at step 14 in a greedy manner using the smallest colour admissible. The process continues at step 16 with updating the degrees of saturation of the uncoloured vertices. Based on the update, a new vertex with the highest degree of saturation will be chosen for colouring and so on. The algorithm runs at step 18 until all the vertices are coloured then the algorithm comes to an end at step 20.

Assume that after graph colouring, we know that the graph is x-colourable, with colours being C, 02 C. And for each colour i (or spectrum slice), the demand is f, respectively, i = 1, ..., x. Here to enable spectrum reuse, f is the maximum demand among all the vertices coloured by colour i. If the total spectrum band available for secondary use is F, then the problem becomes how to fairly allocate (divide) F to x spectrum slices and in the meantime meet their demand? To this end we propose to use the notion of max-mm fairness and proportional fairness.

The problem is to divide a scarce resource among a set of users, each of whom has an equal right to the resource, but some of whom intrinsically demand fewer resources than others. A sharing technique widely used in practice is called max-mm fair share, and is described in a paper by 0. Bertsekas and R. Gallager, Data Networks, Prentice Hall, 1992, the whole contents of which is incorporated herein by reference. Intuitively, a fair share allocates a user with a "small" demand what it wants, and evenly distributes unused resources to the "big" users. Formally, we define max-mm fair share allocation to be as follows: * Resources are allocated in order of increasing demand * No source gets a resource share larger than its demand * Sources with unsatisfied demands get an equal share of the resource This formal definition corresponds to the following operational definition, or an algorithm called progressive filling. Consider a set of sources 1 x that have resource demands f1, 12, ..., f. Without loss of generality, order the source demands so that f1 <= f2 <= ... <= f. Let the server have a capacity F. Then, we initially give Fix of the resource to the source with the smallest demand, f1. This may be more than what source 1 wants, so that F/x -f1 of the resource is still available as unused excess. We distribute this excess evenly to the remaining x-1 sources, so that each of them gets F/x + (FIx-f1)/(x-1). This may be larger than what f2 wants, so we can continue the process. The process ends when each source gets no more than what it asks for, and, if its demand was not satisfied, no less than what any other source with a higher index got. Such an allocation is a max-mm fair allocation, because it maximizes the minimum share of a source whose demand is not fully satisfied.

There are other fairness criteria in resource allocation. Most notably, proportional fair (PF). It has been shown that PF involves maximization of the sum of logarithmic utility functions of individual sources. This is described in a paper entitled "Charging and rate control for elastic traffic', European Transactions on Telecommunications, 1997, by F. Kelly, the whole contents of which are incorporated herein by way of reference.

Another definition of PF in the context of wireless networks is to assign each data flow a data rate or a scheduling priority (depending on the implementation) that is inversely proportional to its anticipated resource consumption. In our spectrum allocation case, we define w1 as the weight associated with spectrum slice i, where w = 1/c, where c as the cost per data bit of using spectrum resource i. Then we replace the demand f1 with (f, I wi), and run the same algorithm as max-mm to obtain the PF allocation. Obviously when the costs are one, PF allocation is the same as max-mm.

The overall spectrum allocation process is summarized in Figure 5. The implementation of the above spectrum allocation method depends on the network architecture in question. For simplicity, we consider a centralized network where a central controller periodically collects all the topology, spectrum and interference information, and is responsible for computing the spectrum allocation and broadcasting the assignment to the nodes. The central controller could be an entity such as the Spectrum Manager defined in 802.22 and the communication protocol to enable spectrum information gathering and allocation is also based on the 802.22 MAC.

Referring to the spectrum allocation flowchart of Figure 5, the first step after the start 22 is the construction of an interference graph, depicted as step 24. The DSATUR algorithm is then applied, at step 26, to the interference graph to derive the chromatic number. A decision is made, shown as step 28, as to whether a max-mm or PF allocation algorithm is then to be applied, implemented at steps 32 and 30 respectively.

As shown in Figure 1, graph colouring with DSATUR yields that the graph is 4-colourable. Assume that the total available spectrum is 100 MHz and located in the spectrum range of 400-500 MHz, and the demands for the four spectrum slices are 20, 26, 40, 50 MHz respectively. To obtain the max-mm fair spectrum allocation, we compute the fair share in several rounds of computation. In the first round, we tentatively divide the resource into four portions of size 25 MHz. Since this is larger than colour l's demand, this leaves 5 MHz left over for the remaining three colours, which we divide evenly among the rest, giving them 80/3 MHz each. This is larger than what colour 2 wants, so we have an excess of 2/3 MHz, which we divide evenly among the remaining two sources, giving them 25 + 5/3 + 1/3 = 27 MHz each. Thus, the max-mm fair allocation is: spectrum slice 1 gets 20 MHz (400-420 MHz), spectrum slice 2 gets 26 MHz (420-446 MHz), spectrum slices 3 and 4 get 27 MHz each (446-473 MHz and 473-500 MHz).

The invention provides a simple method for spectrum allocation in future cognitive radio networks. Although graph colouring is generally NP-hard, the DSATUR heuristic has a polynomial time complexity. The DSATUR algorithm extensively uses various graphs and it is very fast and effective. For example, for a graph with 95 vertices and 755 edges, it just took less than 1 second. It should also be noted that for most graphs, DSATUR yields optimal colouring.

Fairness is an important measure in network resource allocation. This IAR proposes spectrum allocation for secondary users with two commonly-used fairness metrics, i.e. max-mm fair and proportional fair.

The control logic of the algorithms can be implemented in software such as device drivers and the radio hardware does not necessarily require changing.

Apart from DSATUR, there are other heuristics available that can be used in our spectrum allocation problem. For example, to obtain a more accurate graph colouring, one can use iterated greedy algorithms following DSATUR to iteratively revise the colouring. Such an algorithm is described as "Iterated greedy graph colouring and the difficult landscape", Technical Report, Univ. of Alberta, 1992, http://web. cs. ualberta. cal-joe/Coloring!, the whole contents of which are incorporated herein by reference.

Another extension of the invention is to consider a 2-D resource diagram (as shown in Figure 6), in which there are spectrum-time blocks available for secondary use in the grid. Thus the different colours yielded by graph colouring would be distinct spectrum-time blocks instead of spectrum slices. These blocks could be of different sizes depending on the bandwidth and duration that a particular transmission requires.

An exemplary representation of a spectrum is shown in figure 6. The available spectrum 100 comprises a wideband spectrum divided into a plurality (four channels C1, C2, C3 and C4 as illustrated) of distinct frequency bands. The frequency bands can be of fixed bandwidth or varying bandwidth. For the purposes of clarity only, each frequency band is shown as comprising a single channel only Cn, where n is an integer, though it will be apparent that a frequency band could equally be split into a plurality of sub-bands in any of the manners known in the art, or any manner yet to be devised, and remain within the scope of the embodiment.

Furthermore, although channels are in this particular embodiment defined in the spectrum by way of frequency, it will be appreciated that channels may be defined in the medium by any suitable means, such as time, code, space, or any combination thereof, given the nature of the medium and the technology implementation.

In figure 6 the dark blocks are available spectrum opportunities. For example, secondary users may make use of C1, channel 1, during the two dark blocks, that is between times t1 and t3.; they may use C2 during times t3 and t4;and they may use of either channels 03 or C4 during times t1 and t2.

Claims (6)

1. CLAIMS: 1. A method of determining spectrum allocation in a wireless network comprising the step of determining the spectrum allocation required for the wireless network by means of a graph colouring algorithm.
2. 2. A method as claimed in claim 1 wherein the wireless network is a cognitive radio network, the method comprising the step of determining the number of different spectrum slices or blocks required for the cognitive radio network based on the graph colouring algorithm.
3. 3. A method as claimed in claim 1 or claim 2 wherein the method comprises generating an interference graph based on at least one network parameter and applying the graph colouring algorithm to said interference graph to obtain the chromatic number of the graph.
4. 4. A method as claimed in claim 3, wherein said at least one network parameter is selected from a group including network topology, traffic flows and interference constraints.
5. 5. A method as claimed in any one of claims 1 to 4, wherein the spectrum allocation is based on a max-mm fair share algorithm.
6. 6. A method as claimed in any one of claims 1 to claim 4, wherein the spectrum allocation is based on a proportional fair share algorithm.

   7, A method as claimed in any one of claims 1 to 6, wherein the graph colouring algorithm comprises the steps of:- (k) initializing the degrees of saturation (DS) of all vertices to zero; (I) selecting an uncoloured vertex of highest DS; (m) colouring the selected vertex in a greedy manner using the smallest colour admissable; (n) updating the DS of the uncoloured vertices neighbouring the one coloured in the step (c) and (o) if all vertices are not coloured returning to step (b) 8. A method as claimed in any one of claims 1 to 7 wherein the graph colouring algorithm is DSATUR 9. A wireless network operable to establish wireless communication between a plurality of wireless communication apparatus in a wireless communications medium defining a plurality of channels, the network comprising channel availability determining means operable to determine whether any of said one or more channels are available for use by one or more of the wireless communication apparatus; channel selection means operable to select for said one or more wireless communication apparatus, at least one channel available for use based on an available capacity of said at least one channel; and wherein the channel availability determining means is operable to determine said at least one channel allocation by means of a graph colouring algorithm.10. A wireless network as claimed in claim 9, wherein the wireless network is a cognitive radio network, and the channel availability determining means is operable to determine the number of different spectrum slices or blocks required for the cognitive radio network based on the graph colouring algorithm.11. A wireless network as claimed in claim 9 or claim 10, wherein the channel availability determining means is operable to generate an interference graph based on at least one network parameter and applying the graph colouring algorithm to said interference graph to obtain the chromatic number of the graph.12. A wireless network as claimed in claim 11, wherein the channel availability determining means is operable to select said at least one network from a group including network topology, traffic flows and interference constraints.13. A wireless network as claimed in any one of claims 9 to 12, wherein the spectrum allocation is based on a max-mm fair share algorithm.14. A wireless network as claimed in any one of claims 9 or claim 13, wherein the spectrum allocation is based on a proportional fair share algorithm.15. A wireless network as claimed in any one of claims 9 to 15, wherein the graph colouring algorithm comprises the steps of:- (p) initializing the degrees of saturation (DS) of all vertices to zero; (q) selecting an uncoloured vertex of highest DS; (r) colouring the selected vertex in a greedy manner using the smallest colour admissable; (s) updating the DS of the uncoloured vertices neighbouring the one coloured in the step (c) and (t) if all vertices are not coloured returning to step (b) 16. A wireless netwrok as claimed in any one of claims 9 to 15 wherein the graph colouring algorithm is DSATUR.17. A storage medium storing computer executable instructions which, when executed on a general purpose computer controlled wireless network, causes the network to become configured to perform the method of any of claims 1 to 8.18. A signal carrying computer receivable information, the information defining computer executable instructions which, when executed on a general purpose computer controlled wireless network, causes the network to become configured to perform the method of any of claims I to 8.19. A storage medium storing computer executable instructions which, when executed on a general purpose computer controlled wireless network, causes the network to become configured to a wireless network as claimed in any one of claims 9 to 16.20. A signal carrying computer receivable information, the information defining computer executable instructions which, when executed on a general purpose computer controlled wireless network, causes the network to become configured to a wireless network as claimed in any one of claims 9 to 16.