final report project name: future of football team name ... · future work, but outside the scope...

EEL 4924 Electrical Engineering Design

(Senior Design)

Final Report

17 April 2011

Project Name: Future of Football

Team Name: Future of Football

Team Members:

Erik Stegman

Kevin Hoffman

Project Abstract: In the future of football, human error will be virtually eliminated. It is a game of inches and all

too frequently these inches are miscalculated, often affecting the outcome of the game. FOF is a total

football tracking system. Through ultrasonic communication, the precise position of the football will be

constantly recorded with a multilateration algorithm. The coordinates will be displayed in real time via 7

segment LEDs running down both sidelines. As the ball moves, a laser line will point to the correct yard

line on the field. Future work involves synchronizing the position with an instant replay system. This

allows a referee to replay the video and determine the exact location of the ball for any given frame. In

addition, the statistical implications of recording all ball trajectories will revolutionize training methods.

University of Florida EEL 4924—Spring 2011 19-Apr-11 Electrical & Computer Engineering

/18 Final Report: Future of Football

Contents

Abstract ………………………………. 1

Table of Contents ………………………………. 2

Introduction ………...…………………….. 3

Features ………………………………. 3

Technical Objectives ……….……………………… 4

Current Technology ………………………………. 7

Components ………………………………. 8

Project Architecture ………………………………. 11

Future Work ………………………………. 15

Bill of Materials ………………………………. 16

Separation of work ……..………………………... 17

Appendix ………………………………. 19

Figures

Figure 1: Football Circuit ………………………… 5

Figure 2: Field Circuit ………………………… 5

Figure 3: Field Design …………………………. 6

Figure 4: Sonar ………………………… 8

Figure 5: Motor ………………………… 8

Figure 6: Laser ………………………… 9

Figure 7: FPGA ………………………… 9

Figure 8: LEDs ………………………… 9

Figure 9: Chips ………………………… 10

Figure 10: 7 segment LED ………………………… 10

Figure 11: Power ………………………… 10

Figure 12: Transmitter operation ………………………….11

Figure 13: Processor operation …………………………12

Figure 14: Processor initialization …………………………13

Figure 15: Separation of work …………………………17

Figure 16: Gantt chart ………………………… 18



Introduction: Anybody who regularly watches football is familiar with the frequency of bad ball spotting.

Attachment to the way things have always been done seems to be the driving force in maintaining the

current methods. However, with increased processing power, it is only a matter of time before the game

will undergo dramatic technological renovation. To ensure fairness, human error must be eliminated as

completely as possible. A football tracking system is inevitable and overdue. The question remains of

what is the most efficient method to achieve this goal.

The purpose of FOF is to function as a prototype for future ball tracking systems. The goal is to

record the location of the ball at all times with pinpoint accuracy and relay this information to any

observers. Feedback is critical if the system is to be used during game play. For this reason, lights along

the sidelines seem to be an efficient method for identifying the location of the ball. This is a temporary

solution until electroluminescent Astroturf is invented. Sideline lights allow for easy spotting of the ball

between plays. The final location of the ball will also be spotted with a laser from above. Current

regulations would not permit the laser to be turned on until the conclusion of the play. When inches

become important, the instant replay system may be used. This system allows a ref to review the play as

is currently done. However, after determining the exact moment when the play ends and then guessing at

the ball’s location as is currently done, the video may be frozen and the corresponding lights on the field

will display the exact location of the ball at that point in history. This will ensure that every game is

decided by the players, and not by the officiating.

Features: FoF is a real time tracking system to serve as a model for implementation in college and

professional football. Its features include:

Minimal hardware inside of the football to operate a transmitter

Ultrasonic-based multilateration to serve as a model for full-scale RF tracking

Multiple receiver locations for redundancy and increased accuracy

Real-time visual feedback for spectators using lasers and LEDs

High-precision results accurate to within 1” (<1%) to eliminate human error

Potential for a new paradigm of statistical record keeping and analysis



Technical Objectives: The main objective of our project is to triangulate the position of a moving football with a high degree

of accuracy using a fully scalable method. The secondary objective is to record this trajectory and

synchronize this data with video recording.

The football should contain as little hardware as possible in order to meet current regulations

An ultrasonic transmitter inside the football will ping intermittently to remove the need for an echo-

type communication between the field and the ball. This pinging will be accomplished by a 555

timer connected to a Schmitt trigger to send digital pulses. The pulse signal will also need to be

amplified to reach the maximum input voltage of 20v for the transmitter. A block diagram is shown

in Figure 1. Potential problems here include passing the signal through the material of the football.

On a full scale model, RF transmission would be necessary but would require significantly higher

clock speeds. In order to replace this with an ultrasonic signal, a hollow ball with a cage structure for

its outer shell will be used to minimize wave obstruction.

Because the football will have internal components, a wireless charging method is desirable for

future work, but outside the scope of this project.

The ultrasonic ping will be received by four stations located at the corners of the field. A geometrical

algorithm will determine the location of the source of the ping in 2 dimensions on the surface of the

field. It will also therefore be able to determine when the ball has crossed out of bounds. The

resulting data will be capable of plotting the 3rd

dimension as well, but will not be used for this

project. A block diagram of the field circuit is shown in figure 2.

As only 3 receivers are necessary for location in 2 dimensions, the 4th

serves as a means of

redundancy. Data points can be generated from any combination of 3 of these stations. This provides

assurance of signal receipt even if 1 station does not receive the ping. In the event that all 4 stations

receive the signal, the 4 data points can be averaged together to increase the accuracy of the output

data.

The determined location of the ball will be represented as an 8 bit digital signal to be output. This

output will then be run through a decoder to illuminate 7 segment LEDs with the ball coordinates

An alternate method of visual feedback involves LEDs placed about 1 inch apart down the sidelines.

In addition to the LEDs, 2 motors will be suspended above the field with an attached line laser

pointing to the field. The motor is equipped with an encoder for fast response and high resolution.

The laser will be illuminated at the end of the play and will point to the last position of the ball.

The field is a scale model of a full football field. The model is able to fit on a table top. Dimensions

are 96” by 42”, mimicking the dimensions of a regulation field. The processing can be adapted to

any size field simply by entering the new field dimensions into the program. A diagram of the field

is shown in figure 3.

In the future a camera module may be set up to record the play. The video would also be recorded

and stored frame by frame. Each frame would be linked to the position data mentioned above. When

a particular frame is selected, the corresponding positional information will also be selected.

When a frame is selected, the light corresponding to the position of the ball during that from will be

illuminated

The location of first down markers and goal line markers may also be stored. When, during replay, a

frame is selected corresponding to a location which achieves a first down or a touchdown, additional

displays and lights and bells and whistles could be used to inform the viewers.



Figure 1: Football circuit

Figure 2: Field Circuit



Figure 3: Field Design



Current Technology:

What is obvious to fans of most professional ball sports is the significant lack of technology

being implemented during game play. Processing power has increased exponentially in recent decades,

but technological developments in football have been stagnant. With the exception of the instant replay

system, the sport has remained virtually unchanged since its advent. Some other systems are in

development, but the sport’s resistance to change has prevented anything from being implemented this

far.

One of the most intricate tracking systems in use today is employed in professional Cricket and

Tennis. The system, developed by UK based Hawk-Eye LTD, uses video processing to track and project

ball trajectories. The statistical implications have revolutionized the sports. During a cricket match,

every play is recorded and mapped. The result is a comprehensive database of every play.

Understanding the outcome of each play based on ball movement has proved invaluable in terms of

training methods. In tennis, the ball is tracked to determine if it has landed out of bounds. This has been

effective in eliminating a certain amount of human error from the game. No longer is a match unfairly

decided by an inaccurate official.

The shortcoming of this system is its visual based processing. While this is effective in games

where the ball never leaves the sight of the camera, it is inapplicable to football. In football, the ball is

often out of sight of the cameras, the fans, and even the officials and other players. In fact, there is

strategy involved in intentionally hiding the ball from sight. This makes any sort of visual tracking

ineffective. What this sport requires is a method of tracking the ball non-visually so a data point is never

missed.

Significant advancement in the field of football technology is currently under development out of

Carnegie Mellon University. Carnegie’s Football Engineering Research Group is working on

revolutionizing all aspects of the game, from smart shoes that record the impact of kicks, to smart

helmets that record information on injury causing collisions. They are working on a football tracking

system as well. This system involves a significant amount of circuitry inside the football. The ball

contains 3-axis gyros and accelerometers. The collected data is processed inside of the football and the

position is transferred remotely to a base station. The flaw in this design is the quantity of circuitry

inside the ball. While the data is highly accurate, the design requires significant alteration of the current

football design. The league has already shown its resistance to change. Significantly altering the football

design is not likely to be well received. Very specific parameters have been determined for the size,

shape, and weight of the ball. These specifications would have to be revamped to implement this system.

For this reason, an ultra compact and lightweight circuit is desirable inside the ball. An RF

transmission system is small enough to be lost inside the ball. The weight would not be significantly

altered. Such a design is more likely to be accepted. The receiving equipment would also be miniature

and easy to add to a field design. These components could even be hidden inside the pilons which are

already located at the corners of the field.



Components:

Ultrasonic transducers – ProWave: Bistatic open-type 400ST/R120

Figure 4: Sonar

Two parameters were of top concern when selecting the ultrasonic components. Those were

effective range and beam angle. The field is intended to be approximately 4 feet wide and 9 feet long,

corresponding to the dimensions of a regulation size NFL field. This gives a maximum diagonal distance

of approximately 10 feet. Since sensors accurate at 10 feet are usually not accurate under half a foot, we

have elected to move the sensors 6-12 inches away from the field. This gives a necessary effective

distance of 0.5 to 10.5 feet. Although Prowave does not publish the effective distance in their datasheet,

we obtained assurance from the manufacturer that the sensors would meet our needs. The placement of

the sensors at the corners of a rectangle means that for maximum efficiency they need to receive signals

from at least a 90 degree span. Figure 5 shows that the beam angle of this product satisfies this

requirement.

DC motor – 19:1 Metal Gearmotor 37Dx52L mm with 64 CPR Encoder

Figure 5: Motor Using a laser to project on a field of 10 feet with 1 inch accuracy requires high precision in

motor position. For this reason, rather than a servo motor, we selected a dc motor with a feedback

encoder. A servo motor takes a PWM signal of 50 Hz, corresponding to a 20 ms period which is too

slow for our purposes. The DC motor pictured in figure 6 has a gear down ratio of 19 to 1with a 64

count encoder. This yields a motor capable of 500 rpm’s with over 1200 counts per revolution. This is

an accuracy of almost ¼ degree. The encoder feedback is in the form of a PWM signal shown above. A

simple count of the pulses will show how far the motor has turned.



Laser - Laser Straight Projective Leveler with Water Leveler

Figure 6: Laser We are using a laser to mark a yard line on the field. For this reason the point lasers are

undesirable. The laser level provides a laser line up to 20 feet in length which is more than sufficient.

This component will be able to project a laser that will span the width of the field.

Processing – Altera Cyclone II EP2C8T144C8 FPGA with EPSC16 SROM

Figure 7: FPGA Our original intent was to use a low power MSP430 in the football and a high speed MSP in the

field circuit to handle the interrupts and a CPLD to perform all the decoding. A 555 timer circuit has

replaced the need for a controller in the football. We are to use the FPGA to receive the data from the

ultrasound receiver directly, eliminating the need for the other controller. The Altera Cyclone II FPGA

is already available.

LEDs – Bright Green T1 (3mm) Right Angle LEDs

Figure 8: LEDs

A 10 foot field with 2% accuracy requires approximately 64 LEDs. For this reason we have

chosen to use 64 LEDs connected to (4) 4 to 16 decoders. If we find our accuracy in processing is better

than this 2% resolution, we may add more LEDs. A large package of the LEDs shown in figure 9 was

available for a great price so we will try to work with those.



IC’s –

Figure 9: Chips

Circuit requirements for this project include a 555 timer capable of greater than 1kHz

frequencies, a Schmitt trigger to digitize the signal and eliminate rise time data anomalies, a voltage

regulator to convert our 12 volt batteries to the 20 volt max handled by the transducers, an amplifier to

increase the 5 volt digital signal to 20 volts, and 4 decoders to use an 8 bit position signal to light up one

of 64 leds. The addition of a MSP430G2211 has replaced the original intent of using a crystal to produce

the 40kHz wave required. It is programmed to output a PWM that does this portion of the football

circuit

7 segment LEDs – HDSP 5723

Figure 10: 7 segment LEDs For simplicity in feedback and error detection, we have chosen to add 2 double 7 segment LED

displays. The coordinate data for the transmitter will be displayed on these LEDs. This is a simple way

to show the output of the chip processing and debug motor and LED strip behavior.

Power – General Brand A27 12v Alkaline Battery

Figure 11: Battery

The football circuit requires a high voltage at a low weight. The 12 volt alkaline batteries shown

in figure 11 are an inch long and weigh less than an ounce. For this reason they were chosen. The field

can use a 5 volt dc power supply from a wall source.



Project Architecture:

Figure 12: Transmission Flow Chart

The internals of the ball are separated into multiple triangular PCB designs controlling a certain

function of the guts of the ball circuit. One handles power regulation and distribution, Pulsing 20% Duty

Wave, 40kHz wave, and modulating the 40kHz signal into the 20% duty cycle wave and amplifying to

20V to be sent to a Ultrasonic Sensor on every triangle

Hardware:

The hardware is quite simple for this project. Separated into multiple system it can be broken up

down very easy. The first system is the ball circuit.

Ball Circuit – Can be broken up into multiple little system. To produce a wave that will burst a 40kHz

signal every 12ms for only 2ms the ball requires the use of a 555 (NE555N) timer in conjunction with a

Schmitt trigger to digitize the signal. A MSP430(G2211) is used to create the 40kHz signal through its

PWM capabilities. A CMOS switch (CD4016BCE) then uses the Schmitt output as it’s control line to

decide when to let the 40kHz wave through. Once the waveform is created it is them run through a

amplifier (LT1630) to 20V and sent to all Ultrasonic Tranducers.

Once the ball circuit is firing there are receivers (Prowave 400R120) on the field picking up the

signal and amplifying it (LT1630) before sending it along a line to the FPGA prep board.

On the FPGA prep board the signal is then amplified once more (LT1630) and rectified to the

FPGA specifications of a signal -0.5V to 4.6V that can be processed.

The FPGA board is the heart of the system. In it includes a Altera Cyclone II EP2C8T144C8

FPGA with EPSC16 SROM to handle all processing in the realm of multilateration. With the FPGA all

signals can be sent to and from the field representation model. Included are controls to handle display of

segmented LEDs, LED distribution, and controls for powered encoded motor driver with laser level

dependent on quality required and time constraints.



Figure 13: Field circuit processing flow chart



Figure 14: Initialization flow chart

Software:

The field circuit is run by a Cyclone II FPGA. The FPGA takes as inputs the signals from the 4

ultrasonic receivers. Its outputs are binary instructions for the motor encoders. All processing will be

done on this chip. Upon startup, the motor encoders will have no memory of previous position, and

therefore need to be initialized. Figure A shows the process of using buttons to align the motors with

their correct starting position. Based on timing of received signals, the program will calculate the

location of the signal transmission. The flow chart for this process is shown above.

The input signals will be in the form of 40kHz square waves. When the signals arrive, they will

be passed through a debounce circuit. The purpose of this circuit is to set an interrupt signal constantly

high when the oscillating signal is received. When all processing is complete, a refresh signal will be

sent back to these blocks to set the output low.

The arrival of these ultrasonic signals will trigger the starting and stopping of timers. The timers

will then display the difference in time between any 2 received signals. This is a saturation timer which

does not reset once it hits its maximum value. Displaying this value will let the program know that a

signal was never received so it should discard this data point. When all 4 timers have stopped, a ready

signal is sent to the next stage to let the next block know that valid data is available for processing. If

any of the timers hit their maximum value, the data is not valid and will not be passed along. The timers

and debounce circuits are then automatically reset to wait for the next signal. The transmitted signal will

take no longer than 9ms to travel the diagonal length of the field. Therefore we can know if a signal was

missed when the timer reads greater than 9ms. The reset will be activated at 10.5ms which will reset all

values in the circuit and await the next received signal. If any signal is missed, the program will simply

wait for the next one and the motor will then be able to catch up with the current position.

In parallel with the timers are 2 bit registers that store the value of which receiver was triggered

first. This lets the position algorithm know the sign of the arrival time difference. T10 = -(T01). Input to

the registers required two individual bits to be converted into a 2 bit vector.

The next stage of the circuit is the logic block. Location of the ball is determined by a method

known as multilateration. This method is commonly used in locating aircraft. This method can

determine a point in 3 dimensional space without knowing the absolute distances from the transmitter to

the receivers. The only input is the difference in arrival times and Cartesian coordinates of the receivers.

The difference in relative distances is determined by multiplying the time difference by the speed of



sound travel. A comprehensive derivation of the equations will be included in the final report and lab

notebook. The output of this algorithm for our purpose is a value for the x and y coordinates of the

transmitter.

The original design used the 4th

receiver for redundancy. The location should be able to be

calculated from just 3 receivers, I think. Using data from 4 receivers would allow a point to still be

calculated even if it was not received by all 4 stations. In the event it was received by all 4, this method

would also supply 4 data points which could be averaged together for more accuracy. Unfortunately,

extreme difficulty ensued reducing these equations. Only certain combinations of equations yielded

correct data output. Rather than lose more time resolving these problems, we have adjusted the circuit to

intake all received signals and compute using all four measurements. This simplifies the algorithm but

potentially reduces accuracy.

For simplicity, we have implemented 7 segment LEDs to display the output coordinates at this

point in the code. This helps to identify problems in the code for the motors.

This 2D point must then be converted into instructions for the motor. We have employed motors

with hall sensor encoders which return a pulse for each click passed. This motor and encoder

combination yields 1216 counts per revolution. To simplify calculations, data points were rounded to the

nearest inch. An issue of concern was the nonlinearity of change in distance as the laser points to

different locations on the field. As the laser points farther toward the edges, a single count of the encoder

will correspond to a larger distance than the same single count at the center of the field, closest to the

point of laser projection. The change in distance per click is a function of the distance squared. In order

to know how many clicks must be passed to move from point A to point B, the integral of this function

is taken. The integral is an arctan function which is extremely difficult to implement in VHDL. For this

reason we chose to create a look-up table. This table contains 1 value for each position between 0 and

the length or width of the field. These values are approximately the values of the integral at the

corresponding positions. Using this table, the code can take the previous position values, the current

position values, look up both of their corresponding values in the table, and then subtract. This gives the

number of clicks the motor will have to move in order to move from the previous position to the current

position.

Once the number of clicks required is determined, the motors are turned on. The hall sensor

feedback is then used as a clock for a down counter. This counter is loaded with the above value. Once

the count reaches zero, a zero flag is set. The setting of this flag signals the motors to turn off. It also

refreshes the entire circuit and the data collection begins again.

The initial design used a 25.175MHz clock for the timers. At this speed, 18 bit vectors are

necessary to account for the maximum time it could take a signal to traverse the field. These 18 bit

vectors required significant memory to be able to process. The resulting code was accurate to within 0.1

inches, but required 3.5 times more space than was available. Since the circuit boards had already been

created, we chose to scale down our existing code rather than scale up the FPGA and recreate the circuit.

To fit in the space available, the clock speed was divided until the time could be represented with 9 data

bits. The result is a significant loss of accuracy, up to about a 4 inch potential margin of error for most of

the field. There is a problem area as the position approaches midfield. Some of the parameters in the

multilateration algorithm become extremely large here. The required truncation does not allow us to

preserve the accuracy of these numbers. The result is potential garbage data. The code has been

modified to ignore the most extreme errors which occur at exactly midfield. The values are simply

replaced with values corresponding to a position 1 inch away from midfield. This behavior can induce a

4 inch error at some points. If the method were changed to move the value 2 inches away, points on one

end of the field would become more accurate, but points at the opposite end could incur an 8 inch error.

This was more accuracy than we were willing to sacrifice.



Future Work:

The Future of Football system is intended to be a fully functional, fully scalable model of a

system to be implemented on a professional field. The positioning algorithm in use can be applied to any

field for any sport. The code was written such that the only alterations needed for mapping on a larger

scale are the modification of the field dimensions programmed into the algorithm.

As described above, a look-up table was implemented to simplify the use of the arctangent

function. If the same laser method is desirable on a large field, either the table would have to be

significantly extended, or the arctangent function would have to be implemented directly. However, it is

likely that any full scale method would use an alternate method of position identification. A simple

method is to extend the LED strip behavior identified above. LEDs could potentially be placed as

frequently as ever quarter inch extending the entire length of the sideline. For this method to be usable,

the LEDs would need to be extremely bright. A significant hardware requirement is also added for this

method. This feedback system also requires a human to visually locate the illuminated LED and attempt

to place the ball as closely as possible to that point. This still leaves the door open for human error. A

more expensive alternative would be multiple line lasers in place of the LEDs. The laser would then

project a line across the field from the sideline. This would reduce potential error, but as stated earlier,

would not be usable during the duration of the play. This system is more applicable in the instant replay

scenario.

The most significant alteration necessary for full size implementation is the signal transmission

method. Ultrasound was used for this project due to its affordability and ease of implementation.

Ultrasound is not appropriate for a full-scale field for a number of reasons. First, the signal cannot pass

through a leather ball. The signal would not be able to travel sufficient distance to cover the field. The

sonar signal would take too long to cover the field even if it could reach. And it is likely that the

ultrasound would not function properly in the midst of crowd noise. For these reasons, RF transmission

is desirable. RF signals will pass through the leather of the football and even through the other players. It

can travel much farther than the length of the field and would not be interfered with by stadium

conditions.

RF signals travel at light speed. This means that to measure the difference in arrival times, the

counters need to be powered by an ultra-high speed clock. To obtain 1 inch accuracy from the difference

of arrival times, a 12GHz clock is necessary. For professional use, more precision is desirable. A clock

in this range is commercially available for around $10,000 and can be assembled for closer to $1000.

The output vectors from these counters would also be very large, requiring significantly more chip real

estate than was used in this project.

The heart of this tracking method is not its application in real time, but in conjunction with the

instant replay system. By recording all real-time data points and synchronizing them with video data, the

location of the ball at any point in history can be determined. For plays that require further review, this

is an invaluable tool.

By recording the trajectory of the ball as it travels, the entire history of the game is recorded.

Outcomes of plays can be mapped to the movement of the ball during the play. Once this data is

formally organized, football statistics will be overhauled. Training methods would begin to be based on

actual historical data, rather than theory.

Lastly, this method is not confined to use in football. Virtually any ball sport can benefit from

tracking technology. Soccer, hockey and basketball are among the most likely candidates that do not

current feature any sort of tracking.



Bill of Materials:

(4) Ultrasonic transducers – ProWave: Bistatic open-type 400ST/R120 (TX/RX pairs) $87

(2) DC motor – 19:1 Metal Gearmotor 37Dx52L mm with 64 CPR Encoder $80

(2) 3A MC33926 Motor Driver Carrier $50

(2) Laser - Laser Straight Projective Leveler with Water Leveler $25

(128) LEDs – Bright Green T1 (3mm) Right Angle LEDs $2

(1) Processing – Altera Cyclone II EP2C8T144C8 FPGA $20

(1) LM2941CT Adjustable output voltage regulator $1

(1) SN74HC14N Schmitt Trigger $1

(1) LM555CN Timer $1

(4) CD4514 Decoder 4 to 16 $2

(1) MC4558CN 20v Operational Amplifier $1

(1) Hollow rubber ball with unobstructed shell $8

(1) 6’x8’ Fake grass field $18

(2) Cans of white spray paint $2

(1) Lumber 2x4 $3

(1) Advanced Circuits PCB $90

(2) MAN6900 7-segment LED displays $0

(7) LT1630 Operational Amplifier $42

(1) MSP430G2211 Microcontroller $1

(1) LM3940 5V – 3.3V Regulator $1

(1) LT1085 5V regulator $1

(1) CD4016BCE CMOS switch $1



Separation of Work:

Figure 15: Separation of Work



Gantt Chart:

Below is the breakdown of work by order and duration of each project component.

Figure 16: Gantt Chart

References:

Carnegie Mellon Football Engineering Research Group

http://www.footballtracking.org/

Multilateration Executive Reference Guide

http://www.multilateration.com/

Prowave Air Ultrasonic Ceramic Transducers

http://www.prowave.com.tw/pdf/T400S12.PDF

Altera Cyclone II Device Family Data Sheet

http://www.altera.com/literature/hb/cyc2/cyc2_cii5v1_01.pdf

Pololu 19:1 Metal Gearmotor with 64 CPR Encoder

http://www.pololu.com/catalog/product/1442

Pololu MC33926 Motor Driver Carrier


http://www.footballtracking.org/

http://www.multilateration.com/

http://www.prowave.com.tw/pdf/T400S12.PDF

http://www.altera.com/literature/hb/cyc2/cyc2_cii5v1_01.pdf



final report project name: future of football team name ... · future work, but outside the scope...

Documents