2023 IB DP Computer Science 模擬試題連答案詳解

During migration, the legacy system may use outdated file structures, obsolete data types (such as two-digit year fields), or proprietary database schemas. If these do not map directly to the modern system's schema, the data must be transformed. This process can lead to data truncation, loss of historical data, or corruption if characters cannot be decoded correctly by the new system.

Award [1 mark] for identifying a valid compatibility issue (e.g., schema difference, data type mismatch, outdated file format, or encoding issues). Award up to [1.78 marks] for explaining how this issue affects the migration process (e.g., requires complex data transformation, risks data loss/corruption, or causes system downtime).

At the beginning of the fetch stage, the address of the next instruction is stored in the PC. This address is then copied into the MAR. While the MAR uses this address to fetch the instruction from RAM via the address bus, the PC is incremented to point to the subsequent instruction in sequence.

Award [1 mark] for clearly stating the function of the PC (storing the address of the next instruction). Award [1 mark] for clearly stating the function of the MAR (storing the active address for the memory bus). Award [0.78 marks] for distinguishing their relationship during the fetch stage (e.g., address is copied from PC to MAR, allowing the PC to increment independently).

In packet switching, the original message is divided into independent packets, each containing a header with the destination address. Routers dynamically calculate the optimal route for each packet. If a network node along the path fails, subsequent packets are automatically routed through alternative active paths. Once all packets reach the destination, they are reordered. If any packet is lost, the receiver requests a retransmission.

Award [1 mark] for explaining that messages are divided into independent packets with routing headers. Award [1 mark] for describing dynamic routing (packets can take different paths to bypass broken nodes). Award [0.78 marks] for describing packet reassembly and error recovery/retransmission at the destination.

A binary search continually halves the search space. The maximum number of comparisons needed is calculated using \(\lfloor \log_2(N) \rfloor + 1\). For \(N = 200\): \(\log_2(200) \approx 7.64\). Therefore, \(\lfloor 7.64 \rfloor + 1 = 7 + 1 = 8\) comparisons. Alternatively, since \(2^7 = 128\) (insufficient) and \(2^8 = 256\) (sufficient to cover 200 elements), the maximum is 8 comparisons.

Award [1 mark] for showing correct mathematical working (e.g., using log formula, halving sequence, or powers of 2). Award [1.78 marks] for the correct answer of 8 comparisons.

A doubly linked list node contains references to both the next and the previous nodes, which allows traversal in both directions and simplifies deletion of nodes. However, this incurs a disadvantage: it requires more memory storage for the additional 'previous' pointers, and the insertion/deletion logic is more complex because more pointers must be updated.

Award [1 mark] for a valid advantage (e.g., bidirectional traversal, more efficient deletion when pointing directly to a node). Award [1 mark] for a valid disadvantage (e.g., increased memory overhead for the extra pointer, more complex node manipulation). Award [0.78 marks] for a clear comparative contrast.

An open-loop control system performs an action based on a predetermined input or timer, without measuring the actual output or making corrections (e.g., a simple microwave timer). A closed-loop control system constantly monitors the output state using a sensor and feeds this data back to the controller to adjust the input and maintain a target state (e.g., a thermostat-controlled heater). The key component is the sensor that provides feedback.

Award [1 mark] for explaining open-loop (predetermined, no feedback). Award [1 mark] for explaining closed-loop (self-correcting, uses output feedback). Award [0.78 marks] for identifying the sensor / feedback mechanism as the enabling component.

The operating system divides the program's logical address space into fixed-size blocks called pages, and divides the physical RAM into matching blocks called frames. When executing a large program, only the active pages are loaded into physical frames. If the CPU references a page not currently in RAM, a page fault is triggered, prompting the OS to swap an inactive frame back to secondary storage (virtual memory) and load the required page into its place.

Award [1 mark] for explaining the concept of dividing memory into pages and frames. Award [1 mark] for explaining that only active pages reside in RAM while others remain on secondary storage. Award [0.78 marks] for describing the process of handling page faults and swapping pages dynamically.

Polymorphism is the OOP principle where objects of different classes can be treated as objects of a common superclass, and can override common methods to exhibit unique behaviors. It supports extensibility by decoupling client code from specific implementations; if a developer adds a new subclass later, it can be integrated seamlessly into the existing system without rewriting or recompiling the existing code that interacts with the superclass interface.

Award [1 mark] for an accurate definition of polymorphism (different forms/behaviors for the same method call). Award [1 mark] for explaining how it relates to overriding or interfaces. Award [0.78 marks] for outlining how it provides extensibility (adding new classes without breaking or changing existing core code).

1. The Program Counter (PC) holds the address of the next instruction, which is copied to the Memory Address Register (MAR). 2. The MAR sends this address along the address bus to RAM. 3. RAM retrieves the instruction at that address and sends it along the data bus to the Memory Data Register (MDR), where it is temporarily stored before decoding.

Award marks as follows (up to a maximum of 2.78 / 3 marks): - 1 mark: for stating that the MAR holds/receives the memory address of the next instruction (from the Program Counter). - 1 mark: for stating that the MAR sends this address to RAM/primary memory via the address bus. - 1 mark: for stating that the MDR receives and temporarily stores the instruction/data retrieved from RAM via the data bus.

(a) Disadvantages of direct changeover: (1) High risk: If the new ERP system fails or has critical bugs, there is no fallback system available, which can immediately halt retail operations. (2) High pressure on staff: Staff must adapt to the new system immediately without any transition period, increasing the likelihood of operational errors during busy hours.

(b) User Acceptance Testing (UAT): (1) UAT should involve representative end-users (e.g., store cashiers and inventory managers) performing real-world tasks (like processing transactions or updating stock) in a simulated test environment. (2) Feedback is collected, and bugs or usability issues are reported back to the development team to be fixed before final deployment, ensuring that the software meets operational requirements and is user-friendly.

(c) Data migration issues: (1) Data format incompatibility: The legacy system may use structured text or an outdated relational schema that does not match the new cloud ERP's database schema. This requires data mapping and transformation. (2) Data loss or corruption risk: Large-scale data transfers can suffer from transmission errors or incomplete transfers, necessitating checksum verification and comprehensive backups before starting the migration.

(d) System documentation components: (1) Step-by-step user manual/guide for daily tasks (e.g., checkout process). (2) FAQ / Troubleshooting guide for common errors. (3) Quick-reference cheat sheet for keyboard shortcuts and emergency procedures.

Part (a): Award up to [4 marks] max.
- Award [1 mark] for identifying a disadvantage and [1 mark] for explaining it, up to 2 times.
- Disadvantages include: High risk of failure with no fallback, extreme pressure on staff, potential loss of business if downtime occurs.

Part (b): Award up to [4 marks] max.
- Award [1 mark] for mentioning the involvement of actual end-users.
- Award [1 mark] for testing under real-world scenarios or workflows.
- Award [1 mark] for mentioning the formal feedback/bug reporting cycle.
- Award [1 mark] for verifying that the system meets user requirements / sign-off.

Part (c): Award up to [4 marks] max.
- Award [1 mark] for identifying a migration issue and [1 mark] for discussing its impact or solution, up to 2 times.
- Issues include: Data format incompatibility/mapping, data validation/cleansing, risk of data loss/corruption, security during transfer.

Part (d): Award up to [3 marks] max.
- Award [1 mark] for each valid documentation component identified (e.g., user manual, quick-reference guide, troubleshooting guide, FAQs, training videos).

(a) Pseudocode:
maxRow = 0
maxSum = -1
loop R from 0 to 9
rowSum = 0
loop C from 0 to 9
rowSum = rowSum + WAREHOUSE[R][C]
end loop
if rowSum > maxSum then
maxSum = rowSum
maxRow = R
end if
end loop
output maxRow

(b) A binary search requires the data to be sorted to function correctly because it relies on dividing the search space in half based on comparison operations. Sorting an unsorted array of size N takes O(N log N) time, which is slower than a single sequential search taking O(N) time. Therefore, sequential search is more efficient and appropriate for a single search on unsorted data.

(c) The algorithm uses nested loops to traverse the grid. If the dimensions of the grid are N by N, the algorithm must visit every cell once, resulting in N * N operations. This gives a time complexity of O(N^2). If the grid size increases to 1000 by 1000, the number of executions of the inner loop body scales quadratically from 100 to 1,000,000 operations, making the time complexity highly dependent on grid dimensions.

Part (a): Award up to [8 marks] max.
- Award [1 mark] for initializing tracking variables (maxRow, maxSum) correctly.
- Award [1 mark] for implementing the outer loop to traverse all 10 rows.
- Award [1 mark] for resetting the rowSum counter for each new row.
- Award [1 mark] for implementing the inner loop to traverse all 10 columns.
- Award [1 mark] for correctly accessing the 2D array elements using index variables.
- Award [1 mark] for correctly accumulating the row sum inside the inner loop.
- Award [1 mark] for implementing the conditional check (rowSum > maxSum) correctly.
- Award [1 mark] for updating maxRow and maxSum and outputting the result at the end.

Part (b): Award up to [3 marks] max.
- Award [1 mark] for stating that binary search requires the array to be sorted.
- Award [1 mark] for explaining that sorting takes additional time/overhead.
- Award [1 mark] for concluding that for a single search, O(N) sequential search is faster than sorting first.

Part (c): Award up to [4 marks] max.
- Award [1 mark] for identifying the current complexity as quadratic or O(N^2).
- Award [1 mark] for explaining that both dimensions (rows and columns) scale together.
- Award [1 mark] for calculating the scale of the change (from 10x10=100 operations to 1000x1000=1,000,000 operations).
- Award [1 mark] for stating that space complexity remains constant O(1) as only a few scalar variables are used.

(a) DHCP Server: Dynamically assigns an IP address, subnet mask, default gateway, and DNS server address to the student's laptop when it connects to the network. DNS Server: Translates user-friendly domain names (e.g., university.edu) into machine-readable IP addresses so that the laptop can establish a TCP connection.

(b) Packet Switching: (1) The file is split into smaller, manageable units called packets, each with a header containing source IP, destination IP, and sequence number. (2) Each packet is routed independently through the network based on the most efficient path available. (3) If a packet is lost or delayed, the receiving laptop identifies the missing sequence number and requests retransmission of that specific packet (using TCP protocols). (4) Once all packets arrive, they are reassembled in the correct order using their sequence numbers.

(c) Security Measures: (1) WPA3 Enterprise Encryption: Encrypts wireless traffic between laptops and access points using strong cryptographic keys, preventing eavesdropping. (2) 802.1X Authentication: Requires students and staff to log in using their unique university credentials via a RADIUS server, rather than a shared passphrase. (3) Network Segmentation (VLANs): Separates student traffic, staff traffic, and critical administrative servers onto different virtual networks to prevent unauthorized access across segments.

Part (a): Award up to [4 marks] max.
- Award [2 marks] for DHCP: [1] for dynamic IP allocation, [1] for configuration parameters (gateway, subnet).
- Award [2 marks] for DNS: [1] for translation/resolution, [1] for mapping domain names to IP addresses.

Part (b): Award up to [5 marks] max.
- Award [1 mark] for stating files are divided into packets containing headers (IP addresses, sequence numbers).
- Award [1 mark] for stating packets travel independently along different paths.
- Award [1 mark] for describing how sequence numbers are used to reassemble the data in the correct order.
- Award [1 mark] for describing detection of missing/damaged packets at the destination.
- Award [1 mark] for explaining the mechanism of retransmission requests (e.g., TCP ACK/NACK).

Part (c): Award up to [6 marks] max.
- Award [1 mark] for identifying a security measure, and [1 mark] for outlining how it protects data, up to 3 times.
- Valid measures include: WPA3 encryption, 802.1X/RADIUS authentication, VLAN segmentation, MAC address filtering, Firewalls/IDS, VPNs for off-campus access.

(a) Step-by-step state of the linked list:
- After Enqueue('Alice', 2): Head -> [ 'Alice', 2, Next -> null ]
- After Enqueue('Bob', 4): Head -> [ 'Bob', 4, Next ] -> [ 'Alice', 2, Next -> null ] (Bob has higher priority)
- After Enqueue('Charlie', 2): Head -> [ 'Bob', 4, Next ] -> [ 'Alice', 2, Next ] -> [ 'Charlie', 2, Next -> null ] (Charlie has same priority as Alice, so he is placed after Alice due to first-come first-served)
- After Dequeue(): Removes the head node ('Bob', 4).
Final List State: Head -> [ 'Alice', 2, Next ] -> [ 'Charlie', 2, Next -> null ]

(b) Pseudocode:
```
newNode = new Node(newName, newPriority)
if head == null then
head = newNode
else if newPriority > head.Priority then
newNode.Next = head
head = newNode
else
current = head
while current.Next != null and current.Next.Priority >= newPriority do
current = current.Next
end while
newNode.Next = current.Next
current.Next = newNode
end if
```

(c) Advantage: Dynamic memory allocation means the linked list does not have a fixed size limit, preventing overflow issues. Disadvantage: Extra memory is consumed to store the pointer addresses ('Next') for each node, and traversing the linked list to find the insertion point takes O(N) time compared to direct indexing.

Part (a): Award up to [5 marks] max.
- Award [1 mark] for showing Alice correctly enqueued at first.
- Award [1 mark] for showing Bob correctly inserted at the head (before Alice).
- Award [1 mark] for showing Charlie correctly placed after Alice (same priority, stable ordering).
- Award [1 mark] for showing Dequeue correctly removes Bob.
- Award [1 mark] for showing correct pointer transitions, Head pointer, and null pointer at the tail.

Part (b): Award up to [8 marks] max.
- Award [1 mark] for creating the new node correctly with fields set.
- Award [1 mark] for checking the empty list boundary condition (head == null).
- Award [1 mark] for checking the 'insert at head' condition (newPriority > head.Priority).
- Award [1 mark] for updating head pointers correctly when inserting at the head.
- Award [1 mark] for initializing a traversal pointer (current = head).
- Award [1 mark] for implementing a loop to traverse nodes with higher or equal priority.
- Award [1 mark] for checking current.Next != null to prevent null pointer exceptions.
- Award [1 mark] for correctly adjusting Next pointers of current and newNode upon insertion.

Part (c): Award up to [2 marks] max.
- Award [1 mark] for a valid advantage (e.g., dynamic size, no resizing overhead).
- Award [1 mark] for a valid disadvantage (e.g., pointer memory overhead, lack of random access, sequential search required for insertion).

(a) Role of Sensors: Transducer devices that measure physical analog signals from the environment (e.g., temperature) and convert them into electrical signals for the microprocessor. Example: Thermistor/Temperature sensor. Role of Actuators: Devices that receive instructions from the microprocessor to perform physical work or change the state of the system. Example: Fan or heater.

(b) Feedback Loop Process: (1) Temperature sensors constantly measure the ambient greenhouse temperature. (2) This analog signal is converted to digital and sent to the controller. (3) The microprocessor compares the input temperature to the set range (20C - 24C). (4) If temperature < 20C, the microprocessor sends a signal to activate the heater actuator. If temperature > 24C, it activates the cooling fan actuator. (5) If it is within range, no action or shutdown of devices occurs. (6) The cycle repeats continuously, forming a closed feedback loop.

(c) Social/Ethical issues: (1) Economic ruin for the farmer: If the crop is ruined because of a temperature control failure, the farmer faces huge financial losses, affecting their livelihood and employees. (2) Environmental waste/Ethical duty: Ruining thousands of heads of lettuce contributes to massive food waste when regional food security might depend on sustainable agricultural output.

(d) Differences: (1) Feedback: Closed-loop systems use feedback from output sensors to adjust inputs; open-loop systems do not use feedback. (2) Autonomy: Closed-loop systems can self-correct dynamically; open-loop systems simply execute predefined actions regardless of actual outcome.

Part (a): Award up to [4 marks] max.
- Award [1 mark] for defining a sensor and [1 mark] for a valid example (e.g., pH sensor, light sensor, temperature probe).
- Award [1 mark] for defining an actuator and [1 mark] for a valid example (e.g., water pump, grow lights, motor-controlled ventilation vents).

Part (b): Award up to [5 marks] max.
- Award [1 mark] for mentioning continuous measurement of temperature by sensors.
- Award [1 mark] for comparison against target thresholds (20C and 24C) by the processor.
- Award [1 mark] for identifying decision branch when temp is below 20C (activate heater).
- Award [1 mark] for identifying decision branch when temp is above 24C (activate fan/cooler).
- Award [1 mark] for explaining that the output actions change the environment, which is measured in the next cycle (feedback loop).

Part (c): Award up to [4 marks] max.
- Award [1 mark] for identifying an issue and [1 mark] for describing its societal/ethical impact, up to 2 times.
- Issues include: Loss of livelihood/income, food security/supply chain issues, environmental waste, safety of workers (if heaters/generators fail unsafely).

Part (d): Award up to [2 marks] max.
- Award [1 mark] for noting the presence/absence of a feedback mechanism.
- Award [1 mark] for noting the difference in self-correction or adaptability.

Part (a):
- Item is the superclass (parent class) and PerishableItem is the subclass (child class).
- This represents an "is-a" relationship where PerishableItem is a specialized type of Item.
- In UML diagrams, inheritance is represented by a solid line with a hollow arrowhead pointing from the subclass (PerishableItem) to the superclass (Item).

Part (b):
```java
public class Item {
private String id;
private double price;
private int quantity;

public Item(String id, double price, int quantity) {
this.id = id;
this.price = price;
this.quantity = quantity;
}

public double calculateTotalValue() {
return this.price * this.quantity;
}
}
```

Part (c):
```java
public class PerishableItem extends Item {
private int expiryDays;

public PerishableItem(String id, double price, int quantity, int expiryDays) {
super(id, price, quantity);
this.expiryDays = expiryDays;
}

@Override
public double calculateTotalValue() {
double standardValue = super.calculateTotalValue();
if (this.expiryDays < 3) {
return standardValue * 0.5;
}
return standardValue;
}
}
```

Part (d):
- Polymorphism is the ability of an object to take on many forms. Subtype polymorphism allows a subclass method to override a superclass method with the same signature.
- When iterating through an Item[] array, the reference type is Item. However, at runtime, the Java Virtual Machine (JVM) uses dynamic binding (late binding) to determine the actual object type.
- If the object in the array is a PerishableItem, the JVM invokes the overridden calculateTotalValue() method of PerishableItem. If it is a standard Item, it executes the superclass method.

Part (a) [3 Marks]
- [1 mark] Correctly identifying Item as the superclass and PerishableItem as the subclass.
- [1 mark] Identifying the relation as an "is-a" relationship.
- [1 mark] Stating that the UML notation is a solid line with a hollow arrowhead pointing from the subclass to the superclass.

Part (b) [4 Marks]
- [1 mark] Correct constructor signature and parameters for Item.
- [1 mark] Correct initialization of all fields using 'this'.
- [1 mark] Correct method signature for calculateTotalValue().
- [1 mark] Correct calculation and return of price * quantity.

Part (c) [5 Marks]
- [1 mark] Correct class header indicating inheritance (extends Item).
- [1 mark] Correct constructor signature including all 4 parameters.
- [1 mark] Correct use of super(id, price, quantity) as the first line of the constructor.
- [1 mark] Correct override of calculateTotalValue() calling super.calculateTotalValue() to get the base total.
- [1 mark] Correct conditional check on expiryDays and applying the 50% discount correctly.

Part (d) [4.25 Marks]
- [1 mark] Definition of polymorphism (the ability of a reference to take on different forms/behaviors).
- [1.25 marks] Explanation of dynamic binding / late binding at runtime.
- [1 mark] Clear distinction between compile-time reference type (Item) and runtime object type.
- [1 mark] Explanation that the JVM invokes the method corresponding to the actual instance type.

Part (a):
- Encapsulation is the bundling of data and methods that operate on that data into a single unit, while restricting direct access to the object's components.
- Declaring attributes as private prevents external classes from modifying the internal state of an Activity object directly, maintaining data integrity.
- Access is provided through public getter methods (e.g., getType(), getCaloriesBurned(), getDuration()) and optionally setter methods if modification is allowed.

Part (b):
```java
public double totalCaloriesForType(String activityType) {
double totalCalories = 0.0;
for (int i = 0; i < this.activityCount; i++) {
if (this.workouts[i] != null && this.workouts[i].getType().equalsIgnoreCase(activityType)) {
totalCalories += this.workouts[i].getCaloriesBurned();
}
}
return totalCalories;
}
```

Part (c):
```java
public Activity getLongestWorkout() {
if (this.activityCount == 0) {
return null;
}

Activity longest = this.workouts[0];
for (int i = 1; i < this.activityCount; i++) {
if (this.workouts[i] != null && this.workouts[i].getDuration() > longest.getDuration()) {
longest = this.workouts[i];
}
}
return longest;
}
```

Part (d):
- Composition is a strong "has-a" relationship where the child object cannot exist independently of the parent object. If the parent is destroyed, all child objects are also destroyed.
- Aggregation is a weaker "has-a" relationship where child objects can exist independently of the parent object.
- In this scenario, composition is more appropriate because a workout Activity is unique to a specific User. If the User profile is deleted, those specific recorded physical workouts have no independent existence in the tracking app and should be destroyed along with the user.

Part (a) [3 Marks]
- [1 mark] Define encapsulation as restricting direct access to object state / bundling data and methods.
- [1 mark] Explain that private prevents unauthorized or accidental modifications, ensuring data integrity.
- [1 mark] State that public getter/accessor methods are used to safely retrieve the values.

Part (b) [5 Marks]
- [1 mark] Correct method header and initial declaration of accumulator variable (initialized to 0.0).
- [1 mark] Correct loop structure iterating up to activityCount.
- [1 mark] Correct safe check for null before accessing array elements.
- [1 mark] Correct comparison using .equalsIgnoreCase() or .equals() on the activity type.
- [1 mark] Correct accumulation of calories using getCaloriesBurned() and returning the final total.

Part (c) [5 Marks]
- [1 mark] Correct check for empty collection (activityCount == 0) and returning null.
- [1 mark] Correctly initializing the longest activity to the first element.
- [1 mark] Correctly looping through the remaining elements of the array.
- [1 mark] Correctly comparing durations using getDuration() on the objects.
- [1 mark] Correctly updating the longest variable and returning it.

Part (d) [3.25 Marks]
- [1 mark] Distinguish between Composition (strong ownership, lifetime bound) and Aggregation (weak relationship, independent lifetime).
- [1.25 marks] Argue why Composition is suitable (a workout activity has no meaning without its specific User).
- [1 mark] Correct application of context (if user is deleted, activities are deleted).

Part (a):
1. An abstract class cannot be instantiated directly using the new operator.
2. It can contain abstract methods (methods without a body) that must be implemented by any concrete subclass.

Part (b):
```java
public class SmartLight extends Device {
private int brightness;

public SmartLight(String id) {
super();
this.id = id;
this.isOn = false;
this.brightness = 0;
}

@Override
public void adjustSetting(int value) {
if (value < 0) {
this.brightness = 0;
} else if (value > 100) {
this.brightness = 100;
} else {
this.brightness = value;
}
}

public int getBrightness() {
return this.brightness;
}
}
```

Part (c):
- The super keyword is used to call the constructor of the parent class (superclass).
- It is a requirement that the superclass constructor must execute before the subclass constructor code runs, ensuring that the inherited state is properly initialized.
- If not explicitly written, Java automatically inserts a call to the default zero-argument superclass constructor super().

Part (d):
- Declaring the collection as Device[] allows for the storage of any object that inherits from Device (such as SmartLight, SmartThermostat, or SmartSpeaker).
- This provides immense extensibility: if a new smart device type (like SmartLock) is introduced in the future, it only needs to inherit from Device and implement adjustSetting().
- The central controller class does not need to be modified or rewritten to handle the new device type, because it operates on the polymorphic Device reference and calls adjustSetting(), which executes the correct subclass implementation at runtime. This reduces coupling and supports software maintenance.

Part (a) [2 Marks]
Award [1 mark] for each characteristic up to [2 marks]:
- Cannot be instantiated.
- Can contain abstract methods (methods without bodies).
- Can contain both abstract and concrete methods.
- Designed to be a superclass.

Part (b) [6 Marks]
- [1 mark] Correct class header public class SmartLight extends Device.
- [1 mark] Correct declaration of brightness private attribute.
- [1 mark] Correct constructor signature and explicit or implicit call to super().
- [1 mark] Correctly initializing id, isOn to false, and brightness to 0.
- [1 mark] Correct implementation of adjustSetting(int value) checking bounds (clamping lower limit to 0 and upper limit to 100).
- [1 mark] Correct assignment of the parameter value to brightness if within limits.

Part (c) [3 Marks]
- [1 mark] Explanation that super calls the superclass constructor.
- [1 mark] Explaining that it must be the first statement in the subclass constructor.
- [1 mark] Stating that it ensures proper initialization of the inherited attributes from the superclass before subclass logic is executed.

Part (d) [5.25 Marks]
- [1 mark] Identify that Device[] allows for heterogeneous storage (holding multiple different subclasses).
- [1.25 marks] Explain that the controller can use polymorphic method calls (adjustSetting()) without knowing the specific subtype.
- [1.5 marks] Define extensibility in this context: adding new devices (like SmartThermostat) without changing the existing controller code.
- [1.5 marks] Discuss reduction in code duplication and lower coupling between the controller and individual device classes.

Part (a):
```java
public class Flight {
private Seat[][] cabin;

public Flight() {
this.cabin = new Seat[30][6];
char[] letters = {'A', 'B', 'C', 'D', 'E', 'F'};

for (int r = 0; r < 30; r++) {
for (int c = 0; c < 6; c++) {
this.cabin[r][c] = new Seat(r + 1, letters[c]);
}
}
}
}
```

Part (b):
```java
public boolean reserveSeat(int row, char letter, String name) {
if (row < 1 || row > 30) {
return false;
}
int colIndex = letter - 'A';
if (colIndex < 0 || colIndex > 5) {
return false;
}

Seat targetSeat = this.cabin[row - 1][colIndex];
if (targetSeat.isReserved()) {
return false;
}

targetSeat.setPassengerName(name);
targetSeat.setReserved(true);
return true;
}
```

Part (c):
```java
public double getOccupancyRate() {
int totalSeats = 30 * 6;
int reservedCount = 0;

for (int r = 0; r < 30; r++) {
for (int c = 0; c < 6; c++) {
if (this.cabin[r][c] != null && this.cabin[r][c].isReserved()) {
reservedCount++;
}
}
}

return ((double) reservedCount / totalSeats) * 100.0;
}
```

Part (d):
- Advantage: Direct and fast access: A 2D array provides O(1) constant-time access to any seat via mathematical row and column indices (e.g., cabin[row - 1][col]), whereas searching an ArrayList might require O(N) sequential search. It also physically models the real-world structure of cabin seating (rows and columns).
- Disadvantage: Inflexible size: The dimensions of a static array are fixed upon creation. If the airline changes the flight to a larger aircraft (e.g., with 40 rows or 8 columns), the array size cannot be dynamically adjusted without instantiating a completely new array and copying the data over.

Part (a) [5 Marks]
- [1 mark] Correctly instantiating the 2D array: this.cabin = new Seat[30][6];
- [1 mark] Nested loop structures spanning exactly 30 rows and 6 columns.
- [1 mark] Map loop index to seat row number correctly (r + 1).
- [1 mark] Map loop index to character 'A' through 'F' correctly (using array or ASCII arithmetic).
- [1 mark] Instantiate each element with new Seat(...).

Part (b) [5 Marks]
- [1 mark] Validate the row bounds (1 to 30).
- [1 mark] Correctly translate character letter to column index (e.g., letter - 'A') and validate bounds.
- [1 mark] Correctly retrieve the targeted Seat object and check if it is reserved (using accessor, e.g., isReserved()).
- [1 mark] Correctly set the passengerName and set isReserved to true.
- [1 mark] Correctly return true on successful booking and false for invalid inputs/reserved seats.

Part (c) [3.25 Marks]
- [1 mark] Correctly iterate through all 180 elements of the 2D array.
- [1 mark] Accumulate count of reserved seats based on checking isReserved() on the object.
- [1.25 marks] Correctly calculate and return the percentage as a floating-point value (using cast (double) to avoid integer division truncation).

Part (d) [3 Marks]
- [1.5 marks] For explaining a valid advantage (e.g., direct indexed access O(1), physical-to-logical conceptual mapping).
- [1.5 marks] For explaining a valid disadvantage (e.g., fixed memory size, cannot handle varying aircraft cabin sizes dynamically, wastes memory if sizes change).

1. Latency: P2P (Wi-Fi Direct) allows direct drone-to-drone transmission, bypassing the routing overhead of cellular towers and central servers. This achieves sub-millisecond response times critical for immediate evasion maneuvers. 5G, while fast, involves transmission to a base station, core network routing, and central coordination, introducing higher round-trip latency (typically 5 to 20ms) which may be too slow for high-speed dynamic avoidance. 2. Reliability: P2P operates independently of external infrastructure, meaning it remains functional even during widespread power or telecom outages. However, its signal can be severely attenuated by concrete structures and electromagnetic interference in cities. 5G provides robust protocols and licensed spectrums with less interference, but creates a single point of failure (the cellular tower or central controller). 3. Safety Trade-offs: For local safety (avoiding a drone right ahead), P2P is superior due to guaranteed direct visibility. For strategic safety (avoiding areas with high traffic congestion or adverse localized micro-climates), 5G is superior as it offers a centralized macro-view of the entire airspace.

Award up to 7.5 marks in total: Up to 2 marks for analyzing the impact of latency on real-time collision avoidance for both P2P and 5G. Up to 2 marks for explaining reliability differences (infrastructure independence of P2P vs. structured robustness of 5G). Up to 2.5 marks for discussing the local safety implications (micro-level reactions vs. macro-level path coordination). Up to 1 mark for technical accuracy and structured, logical comparison of the two paradigms.

1. Ethical Implications: Continuous surveillance in public and private spaces (e.g., drones flying near residential windows) violates spatial privacy. Citizens cannot give active consent to being filmed by delivery drones, raising ethical dilemmas regarding the commodification of public spaces. 2. Security Implications: Storing raw high-resolution video logs creates a massive data liability. If a drone is physically captured or its storage systems are breached remotely, unauthorized parties could access historical recordings of citizens, security setups of buildings, and daily routines of individuals. 3. Mitigation Evaluation (Edge-Processing): Edge-processing solves the core security issue by transforming raw video into anonymous mathematical vectors (e.g., depth maps, object bounding boxes) directly on the drone's hardware. The raw video is discarded from volatile memory (RAM) instantly, meaning no personal data is written to non-volatile storage or transmitted over the network. 4. Limitations of Mitigation: While highly effective for privacy, edge-processing increases the weight, cost, and power consumption of the drone due to the need for powerful processor chips. This reduces flight times and payload capacity, representing a classic trade-off between privacy compliance and operational efficiency.

Award up to 7.5 marks in total: Up to 2 marks for explaining the ethical implications of continuous urban drone surveillance. Up to 2 marks for identifying and explaining the security risks associated with storing raw high-resolution video footage. Up to 2.5 marks for evaluating the effectiveness of onboard edge-processing as a mitigation strategy, including its privacy benefits. Up to 1 mark for recognizing the technical/physical trade-offs of edge-processing (computational overhead, power usage, drone weight).

1. Dijkstra's Search Space: Dijkstra expands radially from the drone's current position, evaluating every potential node in the urban grid. In a 3D dynamic environment, this leads to an exponential explosion of nodes to process, causing high memory utilization and CPU stress, which drains the drone's battery and delays real-time maneuvering. 2. A* and Heuristics: A* optimizes the search by adding a heuristic function, \(f(n) = g(n) + h(n)\), where \(h(n)\) estimates the cost from node \(n\) to the final delivery point. This restricts the search to a narrow beam oriented toward the destination, resulting in faster execution times and minimal memory footprint. 3. Constraint Analysis: On power-constrained drone hardware, CPU execution time translates directly to power consumption. A* is highly preferred because it minimizes active computation cycles. However, if the dynamic obstacle environment changes constantly, calculating a complex heuristic for many moving objects can introduce its own processing overhead. Additionally, if the heuristic is not admissible (i.e., it overestimates the actual distance), A* may yield a sub-optimal, longer flight path, which also wastes battery in the long run.

Award up to 7.5 marks in total: Up to 2 marks for describing how Dijkstra's algorithm operates (radial, unguided expansion) and its resulting computational impact on resource-constrained systems. Up to 2 marks for explaining how the A* algorithm utilizes a heuristic function \(f(n) = g(n) + h(n)\) to guide the search space. Up to 2.5 marks for analyzing the direct trade-offs on drone hardware (power consumption, processing speed, memory limits, and the risk of sub-optimality due to non-admissible heuristics). Up to 1 mark for precise technical comparison and use of algorithmic terms (search space, heuristic, admissibility, computational complexity).

1. System Latency: Centralized databases execute ACID-compliant transactions in milliseconds, allowing rapid status updates during the critical seconds a drone aligns with an AGV. DLTs require distributed consensus, which takes seconds or minutes to confirm blocks. This delay can stall physical handoff actions or lead to desynchronization between the physical position of the cargo and its digital record. 2. Data Immutability and Security: Once written to a blockchain, the handoff record is cryptographically secured and nearly impossible to alter or delete. This prevents bad actors (or software bugs) from retroactively altering the transaction log to hide package damage or theft. A centralized database lacks this native immutability; anyone with DBA (Database Administrator) privileges can modify logs, making it less secure against insider threats or targeted hacks. 3. Operational Trust: In scenarios where EcoFly collaborates with third-party logistics networks, a shared, permissioned DLT acts as a single, trusted source of truth without requiring any party to trust another's internal IT systems. With a centralized database, the hosting party holds all power, forcing other participants to trust their records blindly. Thus, DLT provides superior multi-organizational trust at the cost of operational speed.

Award up to 7.5 marks in total: Up to 2 marks for evaluating the impact of system latency on real-time physical handoffs (centralized DB speed vs. DLT consensus delays). Up to 2 marks for comparing data immutability and security risks (centralized vulnerability to DBA manipulation vs. DLT cryptographic sequencing). Up to 2.5 marks for discussing operational trust in multi-party logistics scenarios (trustless shared state vs. single-owner control). Up to 1 mark for providing a balanced, clear conclusion or recommendation based on the physical requirements of the handoff system.