AQA A Level Computer Science Paper 2 — Last-Minute Revision Guide (7517/2)

AQA A Level Computer Science Paper 2 (7517/2) is a 2 hour 30 minute written exam worth 40% of the A Level, covering sections 14–21 of the specification: data representation, computer systems, organisation and architecture, consequences of computing, communication and networking, databases, Big Data, and functional programming.

Every question is compulsory, so you can’t skip a topic you’re unsure of. This guide compresses the whole paper into the definitions, model phrases and calculation methods that pick up marks fastest.

How to use this guide before Wednesday: first pass — learn the bold definitions and exam phrases. Second pass — cover the answer and say it aloud. Third pass — do one calculation from each section below.

---

1. Data Representation (Section 14)

Number systems

Term	What to know
Bit	A single binary digit, 0 or 1
Nibble	4 bits — one hex digit
Byte	8 bits
Binary → denary	Add the place values where the bit is 1. E.g. 10110110 = 128+32+16+4+2 = 182
Denary → binary	Use the largest place value that fits, subtract, repeat. E.g. 77 = 64+8+4+1 = 01001101
Binary ↔ hex	Split binary into groups of 4 from the right, convert each to one hex digit. E.g. 1101 0110 = D6

Common mistakes: forgetting leading zeros on 8/16-bit answers; treating A–F as letters instead of values (A=10 … F=15).

Units

• 1 byte = 8 bits
• 1 KiB = 1024 bytes, 1 MiB = 1024 KiB, 1 GiB = 1024 MiB — binary prefixes, powers of 2
• 1 kB = 1000 bytes, 1 MB = 1000 kB — decimal prefixes, powers of 10
• Read the question carefully: KiB/MiB vs kB/MB changes your answer

Two’s complement

• The most significant bit has a negative place value (in 8 bits: -128, 64, 32, 16, 8, 4, 2, 1)
• To negate: write the positive binary, invert every bit, add 1. E.g. +5 = 00000101 → invert 11111010 → +1 = 11111011 = -5
• Overflow = result too large/small for the available bits

Floating point

• Mantissa = significant digits; exponent = how far the binary point shifts
• Normalisation maximises precision by avoiding unnecessary leading bits
• “Increasing the number of mantissa bits increases precision because more significant bits can be stored.”
• “Increasing the number of exponent bits increases range because the binary point can be shifted further.”

Characters, images and sound

• ASCII (7-bit) is mainly for English; Unicode can represent many languages and symbols — “Unicode is more suitable because it can represent a much wider range of characters, including non-English characters and symbols.”
• Bitmap image size (bits) = width × height × colour depth. E.g. 800×600 image at 24-bit colour = 800 × 600 × 24 = 11,520,000 bits = 1,440,000 bytes
• Sound file size (bits) = sample rate × bit depth × channels × duration

Compression, encryption and hashing

• Lossless restores the original exactly (program/text files); lossy permanently removes data (images/sound/video)
• Encryption is reversible with a key (symmetric = same key; asymmetric = public/private key pair). Hashing produces a fixed-length digest and is intended to be one-way — “A hash is used to check integrity because changing the original data should produce a different hash value.”

---

2. Computer Systems (Section 15)

• System software manages the computer (OS, utilities, translators); application software helps the user do tasks
• Operating system roles: processor management, memory management, file management, device management, user interface, security — name the specific role, don’t just say “it runs the computer”
• Compiler translates the whole program before execution (faster to run, errors reported before running); interpreter translates and executes line by line (slower, errors found during execution)

Boolean algebra

Gate	Rule
AND	1 only if both inputs are 1
OR	1 if at least one input is 1
NOT	inverts the input
XOR	1 if inputs differ
NAND / NOR	inverse of AND / OR

• De Morgan’s laws: NOT(A AND B) = NOT A OR NOT B; NOT(A OR B) = NOT A AND NOT B
• Karnaugh maps: use Gray-code ordering (00, 01, 11, 10), group 1s into rectangles sized as powers of 2 (1, 2, 4, 8…), groups can wrap around edges, choose the largest groups

---

3. Organisation and Architecture (Section 16)

Registers — know the full names

Register	Full name	Purpose
PC	Program Counter	Address of the next instruction
MAR	Memory Address Register	Address currently being accessed
MDR	Memory Data Register	Data being transferred to/from memory
CIR	Current Instruction Register	The instruction being decoded/executed
ACC	Accumulator	Results of ALU operations

Fetch-Decode-Execute cycle

1. Address in PC copied to MAR
2. Instruction at that address fetched into MDR
3. Instruction copied to CIR
4. PC incremented
5. Control unit decodes the instruction
6. Instruction executed (using ALU, registers, memory or I/O as needed)

Exam tip: name the registers by their full names — AQA rewards accurate register use heavily on “describe the FDE cycle” questions.

Buses, memory and performance

• Address bus (CPU → memory, carries addresses), data bus (bidirectional, carries data/instructions), control bus (read/write/clock signals)
• “Increasing the width of the address bus allows more memory locations to be addressed.”
• “Increasing the width of the data bus allows more data to be transferred at once.”
• Virtual memory uses secondary storage as extra RAM — slower because disk access is far slower than RAM; thrashing = spending more time swapping pages than executing
• Pipelining overlaps instruction stages; parallel processing uses multiple cores/processors

RISC vs CISC

	RISC	CISC
Instruction set	Small, simple	Large, complex
Instruction size	Fixed	Variable
Pipelining	Easier	Harder
RAM use	More (more instructions needed)	Less (one instruction does more)

Storage

• SSD: fast, robust, no moving parts — but more expensive per GB and limited write cycles (it does have disadvantages!)
• HDD: large capacity, cheap per GB — but slower, moving parts
• Access time ≠ capacity — don’t confuse them

---

4. Consequences of Computing (Section 17)

This section is marked on balanced judgement, not just definitions.

Answer structure:

1. Name the stakeholder (user, company, employee, government, patient, society…)
2. Make a clear point — “This could improve safety because…”
3. Explain the consequence — “This may reduce privacy because…”
4. Give the other side of the argument
5. Finish with a justified judgement if asked to evaluate

Model phrases:

• “One legal issue is data protection, because personal data must be processed fairly and securely.”
• “One ethical issue is bias, because the system may produce unfair outcomes if the training data is not representative.”
• “A benefit for the organisation is reduced cost, but this may disadvantage employees if jobs are automated.”
• “Overall, I would recommend this only if safeguards such as encryption, access controls and human review are used.”

---

5. Communication and Networking (Section 18)

Key terms

• Bandwidth = capacity/rate (don’t confuse with latency); latency = delay before data starts arriving
• LAN (small area), WAN (large area, third-party infrastructure), PAN (around one person)
• Client-server: central management/security, but server failure affects many clients
• Peer-to-peer: no central server, cheaper for small networks, harder to manage consistently

Packets and protocols

• A packet has a header (source, destination, sequence number), payload (the data) and sometimes a trailer (error checking)
• Packet switching: a message is split into packets that may take different routes and are reassembled at the destination — shares links efficiently and routes around failures
• Router forwards packets between networks using routing tables

The layered model

Layer	Job	Example protocols
Application	Network services for applications	HTTP/HTTPS, SMTP, IMAP/POP3, DNS, FTP
Transport	End-to-end communication	TCP (reliable), UDP (fast, less reliable)
Network/Internet	Addressing and routing	IP
Link	Local physical network	Ethernet, Wi-Fi

• “Layering reduces complexity because each layer only needs to interact with adjacent layers.”
• “TCP is suitable when reliability is important because packets can be acknowledged and retransmitted.”
• “UDP is suitable when low latency is more important than perfect reliability.”

Security

• Firewall filters traffic by rules; proxy server sits between client and internet (can cache/filter); VPN creates an encrypted tunnel; WPA/WPA2/WPA3 secure Wi-Fi; TLS/SSL secures communication (e.g. HTTPS)

---

6. Databases (Section 19)

Core vocabulary

• Entity (e.g. Student), attribute (e.g. StudentID), table/relation, record/tuple (a row), field (one item of data)
• Primary key uniquely identifies a record; foreign key links to a primary key in another table; composite key = primary key made of multiple fields
• Referential integrity: foreign key values must match an existing primary key (or be null)

Relationships

• “This is a one-to-many relationship because one customer can place many orders, but each order belongs to one customer.”
• Many-to-many relationships are resolved with a link table containing foreign keys to both original tables

Normalisation

Stage	What it removes
1NF	Repeating groups — values must be atomic
2NF	Partial dependencies on part of a composite key
3NF	Dependencies between non-key attributes

“Normalisation reduces duplicated data and helps avoid insert, update and delete anomalies.”

SQL crash sheet

-- Basic selection
SELECT Forename, Surname
FROM Student
WHERE YearGroup = 13
ORDER BY Surname ASC;

-- Join
SELECT Student.Forename, Student.Surname, Subject.SubjectName
FROM Student
INNER JOIN SubjectChoice ON Student.StudentID = SubjectChoice.StudentID
INNER JOIN Subject ON SubjectChoice.SubjectID = Subject.SubjectID;

-- Aggregate (note: HAVING, not WHERE, after GROUP BY)
SELECT Department, COUNT(*) AS NumberOfStaff
FROM Staff
GROUP BY Department
HAVING COUNT(*) > 5;

ACID and transactions

• Atomicity — all of a transaction happens, or none of it does (“if one part of the bank transfer fails, the whole transaction is rolled back”)
• Consistency — database moves between valid states
• Isolation — concurrent transactions don’t interfere
• Durability — committed changes survive a power failure
• Record locking prevents two users editing the same data simultaneously

---

7. Big Data (Section 20)

The 3 (or 5) Vs: Volume (huge amounts), Velocity (arrives/changes fast), Variety (many formats), plus Veracity (quality/trust) and Value (useful insight).

• Structured data fits a predictable schema (relational tables); unstructured data doesn’t (emails, images, video, social media)
• Distributed processing/storage: work and data split across many machines because one machine isn’t enough
• MapReduce: map — process chunks in parallel; shuffle/sort — group related results; reduce — combine into a final answer (e.g. counting words across millions of documents)
• NoSQL: non-relational, useful for varied/large-scale data — “A NoSQL database may be suitable because the data is varied and does not fit neatly into fixed relational tables.”

---

8. Functional Programming (Section 21)

• Pure function: no side effects, same input → same output always
• Side effect: anything that changes state outside the function (global variables, file writes)
• Immutability: data can’t be changed after creation
• First-class function: can be stored in a variable, passed as an argument, returned from another function
• Higher-order function: takes a function as an argument or returns one
• Recursion: must have a base case, a recursive case, and each call must move closer to the base case

numbers = [1, 2, 3, 4]

# map: apply a function to every item
squares = list(map(lambda x: x * x, numbers))        # [1, 4, 9, 16]

# filter: keep items passing a test
evens = list(filter(lambda x: x % 2 == 0, numbers))  # [2, 4]

# reduce: combine into a single value
from functools import reduce
total = reduce(lambda a, b: a + b, numbers)          # 10

• “map is suitable because the same operation must be applied to every item in the list.”
• “filter is suitable because only values meeting a condition should be kept.”
• “reduce is suitable because the list must be combined into a single value.”

---

Calculation cheat sheet

Question type	Method
Binary → denary	Add place values where the bit is 1
Denary → binary	Use place values largest to smallest
Binary → hex	Group bits in 4s from the right
Two’s complement negative	Negative leftmost place value, or invert + 1
Image size	width × height × colour depth
Sound size	sample rate × bit depth × channels × duration
Transmission time	file size ÷ data rate (check units match!)
Addressable memory	2^(number of address bits)

---

Exam survival checklist

• Underline the command word: state, describe, explain, compare, evaluate, calculate
• Circle the context: school, hospital, business, network, database, image, sound, Big Data
• For explain questions, use “because” in every point
• For compare questions, write about both things in the same sentence/row
• For calculations, write the formula, substitute the numbers, then give units
• For extended answers: benefit → drawback → stakeholder → judgement

Sentence starters to fall back on: “This is because…”, “A benefit of this is…”, “A drawback is…”, “For example…”, “This would be suitable when…”

Good luck for Wednesday! For more topic-by-topic notes and practice, visit the AQA A Level Computer Science revision hub or browse the full A Level revision notes on CompSci Tutoring.