Home > Dynamic Spectrum Allocation

REGUIATORY INTEREST IN DYNAMIC SPECTRUM ALLOCATION :

Spectrum is a national resource, and consequently national regulators have the responsibility to ensure economical use of a society’s spectrum resources. The governments coordinate the development and standardization of radio communication networks and services within the International Telecommunications Union - Radio communications Standardization Sector (ITU-R). At present, the Radio Regulations’ frequency allocations differ depending on the region (Europe, North America, Asia, etc.), and are segmented into precisely defined categories of services (e.g., fixed, mobile, broadcasting, and radiolocation). The convergence of services and growing development of applications based on a combination of two or three
services among fixed, mobile, and broadcasting, raises the issue of simplifying the regulatory framework to adapt to this new situation . At the World Radio Conference (WRC) 2000, Resolution 737, “Review of Spectrum and Regulatory Requirements to Facilitate Worldwide Harmonization of Emerging Terrestrial Wireless Interactive Multimedia (TWIM) Applications,” invited the ITU-R to evaluate the necessity to identify spectrum for TWIM, and to review regulatory methods and service definitions. In light of Resolution 737 an ITL-R Joint Task Group was formed and defined the scope of TWIM as “applications in one or more of the Mobile, Fixed and Broadcasting Services that are capable of supporting hi-directional exchange of information of more than one type (e.g., video, image, data, voice, sound, graphics) between users or between users and hosts.” TWIM was discussed at WRC 2003 and suggested as an agenda item for WRC 2010; further studies will be done until that time. There is also a new resolution for “options to improve the international spectrum regulatory framework” from WRC 2003, which proposes “to examine the effectiveness, appropriateness and impact of the Radio Regulations, with respect to the evolution of existing, emerging and future applications, systems and technologies, and to identify options for improvements in the Radio Regulations.’’ This demonstrates international awareness that thc convergence of radio systems and services will have a significant impact on the way spectrum is regulated. Discussions have also started on a national level. For example, an independent review of spectrum management commissioned by the U. K. government states that “spectrum trading should be implemented in the UK as soon as possible.” In addition, it says “broadcasters should be given the ability to lease spectrum to other uses and/or users.” Outside Europe, the U.S. Federal Communication Commission (FCC) Spectrum Policy Task Force has taken a proactive role in recognizing the potential for new regulation mechanisms to be an important mechanism to allow the modernization of spectrum engineering practices to improve spectrum efficiency. A report from the Spectrum Policy Task Force states that “preliminary data and general observations indicate that many portions
of the radio spectrum are not in use for significant periods of time, and that spectrum use of these ‘white spaces’ (both temporal and geographic) can be increased significantly.” These regulatory developments show that there is a perceived need to bring regulations up to date, and these fit well with the DSA concepts presented here. However, significant changes in regulations would still be required in order to roll out even simple DSA schemes.
World Radio Conferences are scheduled only every three to four years with the scope of the agenda established four to six years in advance. Hence, the regulatory process is slow and not tailored to cope with rapidly changing needs. For example, the 3G spectrum had to be allocated more than 10 years ago, when it was still unclear what 3G would be.

METHODS FOR DYNAMIC SPECTRUM ALLOCATION :

The term dynamic spectrum allocation can potentially cover a range of different subject areas. Several established research fields are related to
DSA, such as dynamic channel allocation (DCA), frequency assignment, unlicensed spectrum access, and spectrum coexistence. If we are considering only a single radio network, the concept of DCA is very close to that of DSA. DCA shares the available radio resources among thc base stations of a radio access network (RAN), and many different schemes have been suggested in this widely researched field. The area of frequency assignment has also been studied for many years . The idea behind frequency assignment is to develop techniques for finding the optimum assignment of frequencies to radio access nodes (e.g., base stations) in order to meet interference and coexistence constraints. However, we are more interested in methods that allow different radio systems with differing characteristics (e.g., broadcast,multicast,unicast, different overlapping cell sizes, various supportable serviced data rates) to dynamically share a set of radio resources between the networks in a composite radio environment scenario. A very relevant and active research area is unlicensed spectrum access. If the spectrum is unlicensed, it is treated as an open resource any conforming device can use. However, this may not be suitable for all scenarios (e.g., where multimedia and delay-sensitive services need to be delivered). Another method of sharing the radio spectrum is to allow different networks to coexist within the same radio spectrum. This is operating successfully today, for example, with the peaceful coexistence of digital and analog TV transmissions in the same hand, and can be extended to consider the coexistence of totally disparate networks through close consideration of their frequency plans . Given that there are schemes for cell-by-cell control and optimization of the radio spectrum for single networks (e.g., DCA and frequency assignment), and also ones that apply little or no control over the systems (e.g., unlicensed access or coexisting systems), we aim to focus on schemes between these two extremes. It must be
remembered that we are specifically looking at schemes that share the spectrum between multiple disparate networks in a composite radio system,
and we aim to share the spectrum between the networks at a RAN level, rather than with cell-by-cell coordination. We therefore concentrate
on methods for permitting two or more networks to share an overall block of spectrum so that spectrum allocations can adapt to either temporal or spatial variations in demand on the networks.
Two schemes to consider are shown in Fig. 1, contiguous and fragmented DSA. These types of scheme have been investigated of scheme have been investigated within the European projects DRiVE, OverDRiVE and TRUST .

Contiguous assignment uses contiguous blocks of spectrum allocated to different RANs, and these are separated by suitable guard hands, much as in fixed assignment. However, the width of the spectrum block assigned to a RAN varies in order to allow for changing demand. This scheme will only allow the spectrum partitioning of a RAN to change at the expense of the spectrally adjacent RANs spectrum. Therefore, if a RAN wishes to increase its allocated spectrum, it will not be able to do so if the spectrally adjacent RAN will not release the spectrum, raising fairness issues. However, this still provides a scheme for allowing spectrum to be used by other RANs if it is not being fully utilized.
The second technique is called fragmented DSA. With this scheme, the spectrum to be dynamically allocated is treated as a single shared block, and any RAN can be assigned an arbitrary piece of spectrum anywhere in this block. This is advantageous if more than two RANs are sharing the spectrum, as the contiguous scheme can be restrictive in these cases. The main disadvantage with this scheme is that it becomes more difficult to control, particularly in terms of interference. This technique can potentially have many guard hands throughout the shared spectrum, and it therefore becomes very important that these be minimized and kept as small as possible, without compromising interference conditions.

OPERATION OF CONTIGUOUS DSA :

As an example of the performance of a DSA scheme, we look at the operation of the contiguous DSA method described above. We consider the scenario of a converged cellular and broadcast system, for example, utilizing the Universal Mobile Telecommunications System (UMTS) 3G cellular network, and the digital video broadcasting terrestrial (DVB-T) network. These, operating in a composite radio system and working together to deliver a mix of unicast and multicast services to users, provides an ideal scenario for DSA operation. The dominant services over each of the networks (voice telephony over UMTS, video over DVB-T) determine the temporal and spatial load demands seen on the networks, and provide the
variations required for DSA.

TEMPORAL DSA :

An example of how a temporal DSA scheme may operate can be seen in Fig. 2. This shows curves of how the loads may change on the different
networks over time, and how this can correspond to changing spectrum partitioning. By comparing the amount of spectrum required for fixed allocation (given hy the sum of the largest demands on the RANs, and labeled "UMTS peak + DVB-T peak") with the amount required for DSA (given by the peak cumulative demand), we can find the extent to which the traffic can be increased with DSA to give the same user satisfaction in the same spectrum as fixed allocation. This gives us an ideal theoretical value for the spectrum efficiency increase. This can be calculated for any set of traffic patterns like those in Fig. 2, and allow for interesting comparisons with algorithms developed to implement temporal DSA. Several implementations of an iterative, realtime, temporal DSA algorithm are possible, hut typical steps in the operation of the temporal DSA algorithm include:

1) Periodic triggering of DSA algorithm: Depending on the traffic patterns, typical timescales could be in the order of tens of minutes to several hours. 2) Management of the trafic on the carriers: This can include packing calls into carriers, locking off carriers from accepting new calls, or selectively dropping calls from the carriers. This ensures that as many carriers as possible can be released for reallocation hy DSA.
3) Prediction of the loads on the networks: Since the DSA algorithm runs periodically, and not on a call-by-call basis, the traffic demands can change significantly during the DSA period. Prediction can be used to allocate spectrum according to the demands that are expected over the whole interval. Prediction schemes can be based on load-histories of past traffic, and timeseries estimation algorithms.
4) Allocation decision: From the predicted loads, combined with the current allocations and the amount of spectrum available for reallocation, an algorithm can decide the allocations that will be applied for this interval. These algorithms have important implications for the performance and fairness of the DSA schemes.

A typical DSA simulated erformance curve can be seen inset in the left curve of Fig. 3, which shows the performance for the overall system of UMTS and DVB-T for both fixed and dynamic assignment with the time varying traffic patterns from Fig. 2. The spectrum efficiency gain of 29 percent is measured as the increase in load supportable at 98 percent user satisfaction. Simulations can be performed for a range of traffic patterns (characterized by the value of the peak cumulative spectrum demand) and compared to the theoretical results. This can be seen in the main graph on the left side of Fig. 3. The comparison highlights several factors that affect the performance of the implemented DSA schemes. These include issues such as the overall amount of spectrum available, and to what degree it is quantized into carriers, the time interval between reallocations, the accuracy of the load prediction, and the readiness of the RANS to free up carriers for reallocation by the DSA algorithm.

SPATIAL DSA :

For the spatial DSA example, we consider the same composite radio system as the temporal case, except we wish to adapt the spectrum allocations
to the regional demands on the networks for a given time. The regional adaptations cannot be performed with arbitrarily fine spatial resolution, and areas of uniform spectrum allocation must be defined, called DSA areas. These should correspond to regions where the traffic demands of the RANs are relatively constant in space. The formation of the DSA areas in this scenario can be seen in Fig. 2, where the size of the area corresponds to a single DVB-T cell. Larger areas are also possible, comprising multiple DVB-T cells. The goal of spatial DSA is to allocate spectrum to RANs according to the traffic demands in each DSA area. Furthermore, it is necessary to coordinate the spectrum allocation between adjacent DSA areas to avoid interference. In particular, the spectrum allocations of different RANs belonging to adjacent DSA areas should not overlap in the same portion of spectrum. In order to avoid this spectrum overlap while still allowing spectrum allocation adaptation to the traffic demand, the guard band needs to be increased to guarantee the coexistence of the different systems, as can be seen by the example allocations shown for the areas in Fig. 2. The extra guard bands represent a critical issue for spatial DSA, since it has to consolidate the need to satisfy the spectrum requirements with the
necessity to limit the guard bands to the minimum possible. The structure of an example spatial DSA scheme can be summarized in three main steps:

1) Calculate the spectrum overlap: In each DSA area this is calculated on the basis of the spectrum requirements of the RANs in its adjacent areas. This operation allows a first estimation of the size of the guard hand in each DSA area, and therefore sets a limit on the portion of allocable spectrum.
2) Perform initial assignment: On the hasis of these limits, calculated for each DSA area, a first spectrum assignment to each RAN can be performed, such that the overall number of satisfied users is maximized.
3) Optimization: After this spectrum assignment, actual spectrum overlap is identified and then eliminated, reducing spectrum allocations where needed.

The right curve of Fig. 3 shows the performance of the overall system of UMTS and DVB-T for both fixed and dynamic assignment. An increase of 14 percent in spectrum efficiency is shown when DSA is used over fixed allocation,
in this example scenario.