- Research Article
- Open Access
OLLAF: A Fine Grained Dynamically Reconfigurable Architecture for OS Support
© S. Garcia and B. Granado. 2009
Received: 15 March 2009
Accepted: 22 September 2009
Published: 11 November 2009
Fine Grained Dynamically Reconfigurable Architecture (FGDRA) offers a flexibility for embedded systems with a great power processing efficiency by exploiting optimizations opportunities at architectural level thanks to their fine configuration granularity. But this increase design complexity that should be abstracted by tools and operating system. In order to have a usable solution, a good inter-overlapping between tools, OS, and platform must exist. In this paper we present OLLAF, an FGDRA specially designed to efficiently support an OS. The studies presented here show the contribution of this architecture in terms of hardware context management and preemption support. Studies presented here show the gain that can be obtained, by using OLLAF instead of a classical FPGA, in terms of context management and preemption overhead.
Many modern applications, for example robots navigation, have a dynamic behavior, but the hardware targets today are still static and this dynamic behavior is managed in software. This management is lowering the computation performances in terms of time and expressivity. To obtain best performances we need a dynamical computing paradigm. This paradigm exists as DRA (Dynamically Reconfigurable Architecture), and some DRA components are already functionals. A DRA component contains several types of resources: logic cells, dedicated routing logic and input/output resources. The logic cells implement functions that may be described by the designer. The routing logic connects the logic cells between them and is also configured by the designer. The I/O resources allow communication outside the reconfigurable area.
Several types of configurable components exist. For example, fine grain architectures such as FPGA (Field Programmable Gate Array) may adapt the functioning and the routing at bit level. Other coarse grain architectures may be adapted by reconfiguring dedicated operators (e.g., multipliers, ALU units, etc.) at coarser level (bit vectors). In a DRA the functioning of the components may change on line during run. FGDRA (Fine Grained Dynamically Reconfigurable Architecture) could obtain very high performances for a great number of algorithms because of its bit level reconfiguration, but this level of reconfiguration induces a great complexity. This complexity makes it hard to use even for an expert and could be abstracted at some level by two ways: at design time by providing design tools and at run time by providing an operating system. This operating system, in order to handle efficiently dynamic applications, has to be able to respond rapidly to events. This can be achieved by providing dedicated services like hardware preemption that lowe configurations and contexts transfer times. In our previous work , we demonstrated that we need to adapt the operating system to an FGDRA, but also we need to modify an FGDRA to have an efficient operating system support.
In this paper we present OLLAF which is an FGDRA specially designed to support dynamics applications and a specific FGDRA operating system.
This paper will be organized as follows. First, an explanation of the problematics of this work is presented in Section 2. Section 3 presents the OLLAF FGDRA architecture and its particularities. In Section 4, an analysis of preemption costs in OLLAF in comparison with others existing platforms, including commercial FPGA using several preemption methods, is presented. Section 5 presents application scenarios and compares context management overhead using OLLAF competing with FPGA, especially the Virtex family. Conclusions are then drawn in Section 6, as well as perspectives on this work.
2. Context and Problematics
Fine Grained Dynamically Reconfigurable Architectures (FGDRA) such as FPGAs, due to their fine reconfiguration grain, allow to take better advantage of optimization opportunities at architectural level. This feature leads in most applications to a better performance/consumption factor compared with other classical architectures. Moreover, the ability to dynamically reconfigure itself at run time allows FGDRA to reach a dynamicity very close to that encountered using microprocessors.
The used model in a microprocessor development gains its efficiency from a great overlapping between platforms, tools, and OS. First between OS and tools, as most main frame OS offer specifically adapted tools to support their API. Also between tools and platform, as an example RISC processors have an instruction set specifically adapted to the output of most compilers. Finally, between platform and OS then, by integrating some OS related component into hardware, MMU is an example of such an overlapping. As for microprocessors, for FGDRAs the keypoint to maximize efficiency of a design model is the inter-overlapping between platforms, tools, and OS.
This article presents a study of our original FGDRA called OLLAF specifically designed to enhance the efficiency of OS services necessary to manage such an architecture. OLLAF has a great inter-overlapping between OS and platform. This particular study mainly focuses on the contribution of this architecture in terms of configuration management overhead compared to other existing FGDRA solutions.
Several studies have been led around FGDRA management that demonstrated the interest of using an operating system to manage such a platform.
Few of them actually propose to bring some modifications to the FGDRA itself in order to enhance the efficiency of some particular services as fast reconfiguration or task relocation. But most of recent studies concentrate on implementing an OS to manage an already existing commercially available FPGA, most often from the Virtex family. This FPGA family is actually the only recent industrial FPGA family to allow partial reconfiguration thanks to an interface called ICAP.
In a previous study, we presented a method allowing to drastically decrease preemption overhead of a FPGA based task, using a Virtex FPGA . In this previous work, as in the one presented here, we made difference between configuration, which relates to the configuration bitstream, and context. Context is the data that have to be saved by the operating system, prior to a preemption, in order to be able to resume the task later without any data loss. In this previous study, we thus proposed a method to manage context, configuration being managed in a traditional way. Conclusions of this study were encouraging but revealed that if we want to go further, we have to work at architecture level. That is why we proposed an architecture called OLLAF  specially designed to answer to problematics related to FGDRA management by an operating system. Among those, we wanted to address problems such as context management and task configuration loading speed, these two features being of primary concern for an efficient preemptive management of the system.
2.2. Related Works
Several researchs have been led in the field of OS for FGDRA [3–6]. All those studies present an OS more or less customized to enable specific FGDRA related services. Example of such services are partial reconfiguration management, hardware task preemption, or hardware task migration. They are all designed on top of a commercial FPGA coupled with a microprocessor. This microprocessor may be a softcore processor, an embedded hardwired core or even an external processor.
Some works have also been published about the design of a specific architecture for dynamical reconfiguration. In  authors discuss about the first multicontext reconfigurable device. This concept has been implemented by NEC on the Dynamically Reconfigurable Logic Engine (DRLE) . At the same period, the concept of Dynamically Programmable Gate Arrays (DPGA) was introduced, it was proposed in  to implement a DPGA in the same die as a classic microprocessor to form one of the first System on Chip (SoC) including dynamically reconfigurable logic. In 1995, Xilinx even applied a patent on multicontext programmable device proposed as an XC4000E FPGA with multiple configuration planes . In , authors study the use of a configuration cache, this feature is provided to lower costly external transfers. This paper shows the advantages of coupling configuration caches, partial reconfiguration and multiple configuration planes.
More recently, in , authors propose to add special material to an FGDRA to support OS services, they worked on top of a classic FPGA. The work presented in this paper try to take advantage of those previous works both about hardware reconfigurable platform and OS for FGDRA.
Our previous work on OS for FGDRA was related to preemption of hardware task on FPGA . For that purpose we have explored the use of a scanpath at task level. In order to accelerate the context transfer, we explore the possibility of using multiple parallel scanpaths. We also provided the Context Management Unit or CMU, which is a small IP that manage the whole process of saving and restoring task contexts.
In that study both the CMU and the scanpath were built to be implemented on top of any available FPGA. This approach showed number of limitations that could be summarized in this way: implementing this kind of OS related material on top of the existing FPGA introduces unacceptable overhead on both the tasks and the OS services. Differently said, most of OS related materials should be as much as possible hardwired inside the FGDRA.
3. OLLAF Architecture Overview
3.1. Specifications of an FGDRA with OS Support
We have designed an FGDRA with OS support following those specifications.
It should first address the problem of the configuration speed of a task. This is one of the primary concerns because if the system spend more time configuring itself than actually running tasks its efficiency will be poor. The configuration speed will thus have a big impact on the scheduling strategy.
In order to enable more choice on scheduling scheme, and to match some real time requirements, our FGDRA platform must also include preemption facilities. For the same reasons as configuration, the speed of context saving and restoring processes will be one of our primary concerns. On this particular point, previous work we have discussed in Section 2 will be adapted and reused.
Scheduling on a classical microprocessor is just a matter of time. The problem is to distribute the computation time between different tasks. In the case of an FGDRA the system must distribute both computation time and computation resources. Scheduling in such a system is then no more a one-dimensional problem, but a three-dimensional one. One dimension is the time and the two others represent the surface of reconfigurable resources. Performing an efficient scheduling at run time for minimizing processing time is then a very hard problem that the FGDRA should help getting close to solve. The primary concern on this subject is to ensure an easy task relocation. For that, the reconfigurable logic core should be splited into several equivalent blocks. This will allow to move a task from one block to any another block, or from a group of blocks to another group of blocks of the same size and the same form factor, without any change on the configuration data. The size of those blocks would be a tradeoff between flexibility and scheduling efficiency.
Another aspect of an operating system is to provide intertask communication services. In our case we will distinguish two cases. First the case of a task running on top of our FGDRA and communicating with another task running on a different computing unit. This last case will not be covered here as this problem concern a whole heterogeneous platform, not only the particular FGDRA computing units. The second case is when two, or more, tasks run on top of the same FGDRA communicate together. This communication channel should remain the same wherever the task is placed on the FGDRA reconfigurable core and whatever the state of those tasks is (running, pending, waiting, ). That means that the FGDRA platform must provide a rationalized communication medium including exchange memories.
The same arguments could also be applied to inputs/outputs. Here again two cases exists; first the case of I/O being a global resource of the whole platform; second the case of special I/O directly bounding to the FGDRA.
3.2. Proposed Solutions
In the center, stands the reconfigurable logic core of the FGDRA. This core is a dual plane, an active plane and a hidden one, organized in columns. Each column can be reconfigured separately and offers the same set of services. A task is mapped on an integer number of columns. This topology as been chosen for two reasons. First, using a partial reconfiguration by column transforms the scheduling problem into a two-dimensional problem (time + 1D surface) which will be easier to handle for minimizing the processing time. Secondly as every column is the same and offers the same set of services, tasks can be moved from one column to another without any change on the configuration data.
In the figure, at the bottom of each column you cannotice two hardware blocks called CMU and HCM. The CMU is an IP able to manage automatically task's context saving and restoring. The HCM standing for Hardware Configuration Manager is pretty much the same but to handle configuration data is also called bitstream. More details about this controller can be found in . On each column a local cache memory named LCM is added. This memory is a first level of cache memory to store contexts and configurations close to the column where it might most probably be required. The internal architecture of the core provides adequate materials to work with CMU and HCM. More about this will be discussed in the next section.
On the right of the figure stands a big block called "HW Sup + HW RTK + CCR". This block contains a hardware supervisor running a custom real time kernel specially adapted to handle FGDRA related OS services and platform level communication services. In our first prototype presented here, this hardware supervisor is a classical 32 bits microprocessor. Along with this hardware supervisor a central memory is provided for OS use only. Basically this memory will store configurations and contexts of every task that may run on the FGDRA. This supervisor communicates with all columns using a dedicated control bus. The hardware supervisor can initiate context transfers, from and to the hidden plane, by writing in CMU's and HCM's registers through this control bus.
Finally, on top of the Figure 1 you can see the application communication medium. This communication medium provides a communication port to each column. Those communication ports will be directly bound to the reconfigurable interconnection matrix of the core. If I/O had to be bound to the FGDRA they would be connected with this communication medium in the same way reconfigurable columns are.
This architecture has been developed as a VHDL model in which the size and number of columns are generic parameters.
3.3. Logic Core Overview
All the material added to support OS in the reconfigurable logic core, concern the configuration memories. That mean that in a user point of view, designing for OLLAF is similar to designing for any common FPGA. This also mean that if we want to improve the functionality of those LE the results presented here will not change.
Configuration data and context data (Flipflops content) constitutes two separate paths. A context swap can be performed without any change in configuration. This can be interesting for checkpointing or when running more than one instance of the same task.
3.4. Configuration, Preemption, and OS Interaction
First, the configuration management service: on the hardware side, each column provides a HCM and a LCM. That means that configurations have to be prefetched in the LCM. The associated service running on the hardware supervisor will thus need to take that into account. This service must manage an intelligent cache to prefetch task configuration on the columns where it might most probably be mapped.
Second, the preemption service: the same principle must be applicable here as those applied for configuration management, except that contexts also have to be saved. The context management service must ensure that there never exists more than one valid context for each task in the entire FGDRA. Contexts must thus be transferred as soon as possible from LCM to the centralized global memory of the hardware supervisor. This service will also have a big impact on the scheduling service as the ability to perform preemption with a very low overhead allows the use of more flexible scheduling algorithms.
Finally the scheduling service, and in particular the space management part of the scheduling: it takes advantage of the column topology and the centralized communication scheme. The reconfigurable resource could then be managed as a virtual infinite space containing an undetermined number of columns. The job is to dynamically map the virtual space into the real space (the actual reconfigurable logic core of the FGDRA).
3.5. Context Management Scheme
In , we proposed a context management scheme based on a scanpath, a local context memory and the CMU. The context management scheme in OLLAF is slightly different in two ways. First, every context management related material is hardwired. Second, we added two more stages in order to even lower preemption overhead and to ensure the consistency of the system.
As context management materials are added at hardware level and no more at task level, it needed to be splited differently. As the programmable logic core is column based, it was natural to implement context management at columns level. A CMU and a LCM have then been added to each column, and one scanpath is provided for each column's set of flipflops.
Contexts are transferred by the CMU into LCM in the hidden plane with a scanpath. Because the context of every column can be transferred in parallel, LCM is placed at column level. It is particularly useful when a task uses more than one column. In the first prototype, those memories can store 3 configurations and 3 contexts. LCM optimizes access to a bigger memory called the Central Context Repository (CCR).
CCR is a large memory space storing the context of each task instance run by the system. LCM should then store context of tasks who are most likely to be the next to be run on the corresponding column.
After a preemption of the corresponding task, a context can be stored in more than one LCM in addition to the copy stored in the CCR. In such situation, care must be taken to ensure the consistency of the task execution. For that purpose, contexts are tagged by the CMU each time a context saving is performed with a version number. The operating system keeps track of this version number and also increments it each time a context saving is performed. In this way the system can then check for the validity of a context before a context restoration. The system must also try to update the context copy in the CCR as short as possible after a context saving is performed with a write-through policy.
In best case, preemption overhead can then be bound to one clock cycle.
A scenario of a typical preemption is presented in Figure 5. In this scenario we consider the case where context and configuration of both tasks are already stored into LCM. Let us consider that a task T1 is preempted to run another task T2, scenario of task preemption is then as follows:
(i)T1 is running and the scheduler decides to preempt it to run T2 instead,
(ii)T2 is configuration and eventual context are shifted on the hidden plane,
(iii)once the transfer is completed the two configuration planes are switched,
(iv)now T2 is running and T1's context can be shifted out to be saved,
(v)T1's context is updated as soon as possible in the CCR.
4. Preemption Cost Analysis
4.1. OLLAF versus Other FPGA Based Works
This section presents an analytic comparison of preemption management efficiency on different solutions using commercial FPGA platform and on our FGDRA OLLAF. The comparison was made on six different management method to transfer the context and the configuration for the preemption incuding the methods in use in OLLAF.
The six considered methods are
a solution based on the Xilinx XAPP290  using ICAP interface to transfer both context and configuration and using the readback bitstream for context extraction,
a solution that is similar to Scan solution but using 8 parallel scanpath as described in  to transfer the context, ICAP interface is still used for configuration transfer,
a solution that uses a dual plane scanpath similar to the one used in OLLAF for context transfer and ICAP for configuration transfer. This method is also studied in , referred as a shadow Scan Chain,
a solution that uses ICAP for configuration transfer and the memory mapped solution proposed in  for context transfer,
a solution that use separate dual plane scanpath for configuration transfer and context transfer as used in the FGDRA architecture proposed in this article.
We defines the preemption overhead as the cost of a preemption for the system in terms of time, expressed as a number of clock cycles or "tclk". In the same way, all transfer times are expressed and estimated in number of clock cycle as we want to focus on the architectural view only. Task sizes will be parameterized as , the number of flipflops used.
Preemption overhead can be due to context transfers (two transfers: one from the previously running task to save it is context and one to the next task to restore it is context), configuration transfers (to configure the next task) and eventually context's data extraction (if the context's data are spreaded among other data as in the XIL solution).
The five first solution uses the ICAP interface as configuration transfer method. Using this method, transfers are made as configuration bitstream. A configuration bistream contains both a configuration and a context. In the same way, for the XIL solution that also use the ICAP interface for context saving, the readback bitsteam contains both a configuration and an context. In this case only context is useful. But we need to transfer both configuration and context and then to spend some extra time to extract the context.
According to , we can estimate that for an flipflop IP, and so an bits context, the configuration is 20 bits. That means a typical ICAP bitstream of 21 bits.
Comparison of task preemption overhead for 713 flipflops task.
Those results show that in this case, using our method leads to a preemption overhead around 500 times smaller than using the best other method.
In OLLAF, both context and configuration transfers are hidden due to the use of dual plane. The latency between the moment a preemption is asked and the moment the new task effectively begins to run can also be studied. This latency only depends on the size of the columns. In the worst case this latency will be far shorter than the OS tick period. OS tick period being in any case the shortest time in which the system must respond to an event, we can consider that this latency will not affect the system at all.
5. Dynamic Applications Cases Studies
In this section, we will consider few applications cases to demonstrate the contribution of the OLLAF architecture especially for the implementation of dynamical applications. Applications will be here presented as a task dependency graph, each task being characterized by its execution time, its size as a number of columns occupied, and eventually its periodicity.
In this study, we consider an OLLAF prototype with four columns. The study consists of comparing the execution of a particular application case using three different context transfer methods. The first considered context transfer method will be the use of an ICAP-like interface, this will be the reference method as it is the one considered in most of today's works on reconfigurable computing. The second consider method will be the method used in the OLLAF architecture as presented earlier. We will here consider using LCM of a size of 3 configurations and 3 contexts. Then in order to study in more detail the contribution of dual planes and of the LCM we will also considered a method consisting on an OLLAF like architecture but using only one plane. As the use of a dual planes will have a major impact on the reconfigurable logic core's performance, this last case is of primary concern to justify this cost.
Applications presented here involves each a first task T1 which has a periodicity of 40 ms, each time execution of this task finish, the remaining sequence begin (creation of task T2 ) and a new instance of T1 is ran. This correspond to a typical real time imaging system, a task is in charge of capturing a picture, then each time a picture as been fully captured, this picture is being processed by a set of other tasks, while the next picture is being captured.
5.1. Considered Cases
This case is particularly interesting for many reasons.
First the loading condition of the task dynamically depends on the previous iteration of the sequence. As an example, if no object has been detected in the previous scene, then no has been created and thus is still fully operational into the active plane, it may only eventually have to be reseted. If now 3 or more objects has been detected and thus all the three free columns has been used, then the full context of have to be loaded from the second plane or in some cases from the local caches.
Another interesting aspect occurs when 4 objects are detected and so 4 are created and must be executed. In that case if three first are executed, one on each free column, and then the fourth is executed on one random column, then a new image will be arrived before processing of the current one is finished, in other terms, the deadline is missed. However, by scheduling those four instances using a simple round robin algorithm with a quantum time of 5 ms, real time treatment can be achieved. It should be noticed that this scheduling is only possible if preemption is allowed by the platform.
Results for case 1 execution.
Results for case 2 execution.
Considering the results using OLLAF, for the first iteration of the sequence, give information about the contribution of the dual planes while the results for the next iterations using "OLLAF simple" give information about the contribution of the LCM only. If we now consider the result for next iterations using OLLAF, we can see that a major gain is obtained by combining LCM and a dual planes. In the cases considered here, this gain is a factor between for case 2 and for case 1 and 3 compared to the ICAP solution.
We also have to consider the scalability of proposed solutions. Transfers at level are not dependent of either column size or number of columns in the considered platform. transfer time depend on the size of each column but not on the number of columns in use. transfers not only depends on the column size but also on the number of column as all transfers at this level share the same source memory (CCR) and the same bus. We can see that using the classical approach will face some scalability issues while OLLAF offer a far better scalability potential as transfers cost is far less dependent on the platform size.
Those results not only prove the benefit of the OLLAF architecture, but they also demonstrate that the use of LCM allows to take better advantage of dual planes.
In this paper we presented a Fine Grained Dynamically Reconfigurable Architecture called OLLAF, specially designed to enhance the efficiency of Operating System's services necessary to its management.
Case study considering several typical applications with different degrees of dynamicity revealed that this architecture permits to obtain a far better efficiency for task loading and execution context saving services than actual FPGA traditionally used as FGDRA in most recent studies. In the best case, task switching can be achieved in just one clock cycle. More realistic statistical analysis showed that for any basic dynamic case considered, the OLLAF platform always outperform commercially available solution by a factor around to concerning contexts transfer costs. The analysis showed that this result can be achieved thinks to the combination of a dual planes and an LCM.
This feature allows fast preemption and thus permit to handle dynamic applications efficiently. This also open the door to lot of different scheduling strategies that cannot be considered using classical architecture.
Future works will be led on the development of an online scheduling service taking into account new possibilities offered by OLLAF. We could include prediction mechanism in this scheduler performing smart configurations and contexts prefetch. Being able to predict in most cases the future task that will run in a particular column will permit to take even better advantage of the context and configuration management scheme proposed in OLLAF.
This work contribute to make FGDRAs a much more realistic option as universal computing resource, and make them one possible solution to keep the evolution of electronic system going in the more than moore fashion. For that purpose, we claim that we have to put a lot of efforts to build a strong consistence between design tools, Operating Systems and platforms.
- Garcia S, Prevotet J, Granado B: Hardware task context management for fine grained dynamically reconfigurable architecture. Proceedings of the Workshop on Design and Architectures for Signal and Image Processing (DASIP '07), November 2007, Grenoble, France Google Scholar
- Garcia S, Granado B: OLLAF: a fine grained dynamically reconfigurable architecture for os support. Proceedings of the Workshop on Design and Architectures for Signal and Image Processing (DASIP '08), November 2008, Grenoble, France Google Scholar
- Simmler H, Levinson L, Männer R: Multitasking on FPGA coprocessors. Proceedings of the 10th International Conference on Field-Programmable Logic and Applications (FPL '00), August 2000, Villach, Austria, Lecture Notes in Computer Science 1896: 121-130.Google Scholar
- Chen G, Kandemir M, Sezer U: Configuration-sensitive process scheduling for FPGA-based computing platforms. Proceedings of the Design, Automation and Test in Europe Conference and Exhibition (DATE '04), February 2004, Paris, France 1: 486-493.View ArticleGoogle Scholar
- Walder H, Platzner M: Reconfigurable hardware operating systems: from design concepts to realizations. Proceedings of the International Conference on Engineering of Reconfigurable Systems and Algorithms (ERSA '03), 2003 284-287.Google Scholar
- Wigley G, Kearney D, Warren D: Introducing reconfigme: an operating system for reconfigurable computing. Proceedings of the 12th International Conference on Field Programmable Logic and Application (FPL '02), September 2002, Montpellier, France 2438: 687-697.Google Scholar
- Ling X-P, Amano H: Wasmii : a data driven computer on virtuel hardware. Proceedings of the IEEE Workshop on FPGAs for Custom Computing Machines, April 1993, Napa, Calif, USA 33-42.View ArticleGoogle Scholar
- Shibata Y, Uno M, Amano H, Furuta K, Fujii T, Motomura M: A virtual hardware system on a dynamically reconfigurable logic device. Proceedings of the IEEE Symposium on FPGAs for Custom Computing Machines (FCCM '00), April 2000, Napa Valley, Calif, USA Google Scholar
- DeHon A: DPGA-coupled microprocessors: commodity ICs for the early 21st century. Proceedings of the IEEE Workshop on FPGAs for Custom Computing Machines (FCCM '94), April 1994, Napa Valley, Calif, USA 31-39.View ArticleGoogle Scholar
- Xilinx : Time multiplexed programmable logic device. US patent no. 5646545, 1997Google Scholar
- Li Z, Compton K, Hauck S: Configuration caching techniques for FPGA. Proceedings of the IEEE Symposium on FPGA for Custom Computing Machines (FCCM '00), April 2000, Napa Valley, Calif, USA Google Scholar
- Nollet V, Coene P, Verkest D, Vernalde S, Lauwereins R: Designing an operating system for a heterogeneous reconfigurable SoC. Proceedings of the 17th International Parallel and Distributed Processing Symposium (IPDPS '03), April 2003, Nice, France 174.Google Scholar
- Xilinx : Two flows for partial reconfiguration: module based or difference based. Xilinx, Application Note, Virtex, Virtex-E, Virtex-II, Virtex-II Pro Families XAPP290 (v1.2), September 2004Google Scholar
- Koch D, Haubelt C, Teich J: Efficient hardware checkpointing: concepts, overhead analysis, and implementation. Proceedings of the 17th International Conference on Field Programmable Logic and Applications (FPL '07), August 2007, Amsterdam, The Netherlands Google Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.