1.0 Internet of Things (IoT) – Core Technologies & Implications
1.1 Fundamentals of IoT
-
Machine-to-Machine (M2M) Communication:
Direct communication between devices without human intervention.
Working Principle: Sensors/actuators → Gateway → Network → Application server.
Example: Smart meter sending usage data to utility server.
-
Practical Applications:
-
Smart Homes (automation, security)
-
Smart Cities (traffic, waste management)
-
Healthcare (remote monitoring)
-
Agriculture (soil, crop monitoring)
-
Industrial IoT (predictive maintenance)
-
-
Key Components of IoT Network:
-
Sensors/Actuators: Data acquisition and action.
-
Connectivity: Protocols (Wi-Fi, Bluetooth, LPWAN).
-
Data Processing: Edge/cloud analytics.
-
User Interface: Dashboards, mobile apps.
-
-
IoT Conceptual and Architectural Framework:
Common three-layer architecture:
-
Perception Layer: Sensors/actuators for physical world interaction.
-
Network Layer: Data transmission via internet/communication networks.
-
Application Layer: User-facing services and analytics.
Extended models: Five-layer (adding middleware and business layers).
-
-
Communication APIs and Their Role:
Enable standardized device-to-cloud/device-to-device communication.
-
RESTful APIs: HTTP-based, stateless, for resource access.
-
MQTT/CoAP: Lightweight, publish-subscribe or request-response for constrained devices.
[!TIP] APIs abstract hardware complexity, allowing interoperability and scalable application development.
-
1.2 Sensor Technologies
-
Sensor Evolution:
From mechanical → electronic → smart sensors (embedded processing).
Significance: Enables real-time, accurate data acquisition for IoT decision-making.
-
Classification of Sensors:
| Type | Definition | Examples | |----------------|-----------------------------------------|-------------------------------| | Scalar | Measures single magnitude | Temperature, pressure sensors | | Vector | Measures magnitude and direction | Accelerometer, gyroscope |
-
Sensor Performance Parameters:
-
Bias: Constant systematic error; output offset from true value.
-
Drift: Output change over time for constant input (e.g., temperature sensor drift).
-
Hysteresis Error: Difference in output for increasing vs. decreasing input (e.g., mechanical sensors).
-
Quantization Error: Error due to analog-to-digital conversion; step size = $$\displaystyle \frac{\text{range}}{2^{\text{bits}}} $$.
-
1.3 IoT Networking Protocols & Standards
-
6LoWPAN:
IPv6 over Low-Power Wireless Personal Area Networks.
Contribution: Adapts IPv6 for low-power devices via header compression and fragmentation.
-
RFID:
Working Principle: Tag (passive/active) → Reader (emits RF) → Antenna → Backend system.
Applications: Inventory tracking, access control, supply chain.
-
Bluetooth in IoT:
Bluetooth Low Energy (BLE): Low-power, short-range (10–100 m), used in wearables, beacons.
-
Wireless Communication Methods:
| Method | Range | Power | IoT Role | |--------------|-----------------|-----------|----------------------------------| | Wi-Fi | ~100 m | High | High-bandwidth local networks | | Cellular (4G/5G) | km | High | Wide-area, mobile IoT | | LPWAN (LoRa) | km | Low | Long-range, low-data-rate |
-
Advanced Message Queuing Protocol (AMQP):
-
Features: Reliable, interoperable, secure messaging.
-
Components: Exchange (routes messages), Queue (stores messages), Binding (links exchange to queue).
-
Frame Types:
-
Method: Defines operations (e.g.,
basic.publish). -
Content: Carries message data.
-
Header: Additional metadata.
-
Heartbeat: Keep-alive.
-
Role: Enables decoupled, asynchronous communication between IoT devices and servers.
-
-
Constrained Application Protocol (CoAP):
-
Description: Lightweight HTTP-like protocol for constrained devices (UDP-based).
-
Message Types:
-
CON (Confirmable): Requires ACK.
-
NON (Non-confirmable): No ACK.
-
ACK (Acknowledgment).
-
RST (Reset): Error indication.
-
-
Request-Response Model:
Client sends CON/NON request → Server responds with ACK + payload (or separate CON).
[!TIP] CoAP uses
GET/PUT/POST/DELETEmethods similar to HTTP but optimized for low-power networks.
-
1.4 IoT Hardware & Platforms
-
Arduino Boards:
Microcontroller-based, easy prototyping, abundant GPIO pins, suitable for simple sensor/actuator control.
-
Raspberry Pi:
Single-board computer, runs Linux, higher processing power, supports complex applications (e.g., image processing).
-
Cloud Computing in IoT:
-
Significance: Provides scalable storage, processing, and device management.
-
Role:
-
Storage: Handle massive IoT data volumes.
-
Processing: Run analytics, machine learning on aggregated data.
-
-
Cloud Storage Models:
| Model | Ownership | Use Case | |-------------|--------------------|----------------------------------| | Public | Third-party (AWS) | Cost-effective, scalable | | Private | Organization-owned | High security, compliance | | Hybrid | Mixed | Balance of cost and control | | Community | Shared by group | Specific industry needs |
-
1.5 IoT Actuation & Integration
-
Types of Actuators:
-
By Motion: Rotary (motors), linear (solenoids).
-
By Energy: Electric, hydraulic, pneumatic.
-
By Control: On/off, proportional, servo.
-
-
Integration of Sensors, Actuators, and Communication Protocols:
Typical Flow:
Sensor data → Communication protocol (e.g., MQTT) → Cloud/edge processing → Decision → Actuator command via protocol.
Example: Temperature sensor (data) → CoAP → Cloud analytics → If > threshold → BLE → Smart plug (actuator) to turn off AC.
2.0 Social Networks & Web Technologies – Legal & Analytical Perspectives
2.1 Foundations of the Social Web
-
Emergence and Evolution:
Web 1.0 (static) → Web 2.0 (user-generated content) → Social Web (platforms like Facebook, Twitter enabling interactions).
-
Types of Web-Based Networks:
-
Online Social Networks (OSNs): Facebook, Instagram.
-
Professional Networks: LinkedIn.
-
Microblogging: Twitter.
-
Collaborative Networks: Wikipedia.
-
-
Importance of Social Network Analysis (SNA):
Understand relationships, influence, information diffusion, detect communities, security threats (e.g., fake profiles).
2.2 Semantic Web & Ontologies
-
Resource Description Framework (RDF):
Data model using triples:
(subject, predicate, object)to describe resources.Example:
(Alice, knows, Bob). -
RDF Schema (RDFS):
Extends RDF with classes and properties (e.g.,
rdfs:Class,rdfs:subClassOf). -
Web Ontology Language (OWL):
Unique Features:
-
Rich expressiveness (class equivalence, property restrictions).
-
Reasoning support (infer new facts).
-
Formal semantics.
-
-
FOAF (Friend of a Friend):
Ontology for social individuals and relationships.
Uses RDF to define
foaf:Person,foaf:knows,foaf:mbox(email), enabling decentralized social graphs.
2.3 Web Community Analysis
-
Measuring Evolution Metrics in Web Communities:
Track changes over time from web archives:
-
Size: Number of nodes/edges.
-
Density: $$\displaystyle \frac{2E}{V(V-1)} $$ for undirected graphs.
-
Diameter: Longest shortest path.
-
Centralization: Variation in node degrees.
-
-
Community Definitions:
-
Local: Subgraph with dense internal connections (e.g., clique).
-
Global: Partition of entire network (e.g., modularity-based).
-
Vertex-Based: Defined by node attributes (e.g., users with similar interests).
-
-
Network Reduction Techniques:
Simplify large networks:
-
Pruning: Remove low-degree nodes.
-
Coarsening: Merge similar nodes.
-
Core-Periphery: Focus on core nodes.
-
2.4 Human Experience & Context
-
Enabling New Human Experiences:
-
Social Networks: Connect people globally, share experiences.
-
Reality Mining: Collect mobile sensor data (location, Bluetooth) to model human behavior.
-
Context Awareness: Systems adapt to user context (location, activity, time) for personalized services.
-
2.5 Privacy & Security in Online Social Networks (OSNs)
-
Privacy Issues:
-
Data leakage (personal info exposed).
-
Profiling and targeted manipulation.
-
Unauthorized access to private posts.
-
-
Attack Spectrum and Countermeasures:
| Attack | Description | Countermeasure | |--------------------------|------------------------------------------|----------------------------------------| | Plain Impersonation | Fake profile pretending to be someone | Profile verification, AI detection | | Profile Cloning | Copying existing profile details | Watermarking, anomaly detection | | Profile Hijacking | Taking over legitimate account | Strong authentication, login alerts | | Profile Porting | Transferring profile to another service | Cross-platform identity verification | | Censorship Attacks | Suppressing content (e.g., via Sybils) | Robust graph algorithms, monitoring |
-
Challenges for Decentralized OSNs (DOSNs):
-
Trust management without central authority.
-
Data consistency across nodes.
-
Scalability and performance.
-
Privacy preservation in distributed storage.
-
2.6 Social Network Analysis Techniques
-
Types of Centrality:
-
Degree: Number of direct connections.
-
Betweenness: Fraction of shortest paths through a node (bridge role).
-
Closeness: Inverse average distance to all nodes.
-
Eigenvector: Influence of neighboring nodes (e.g., PageRank).
-
-
Clustering Methods:
-
Hierarchical: Agglomerative (bottom-up) or divisive (top-down).
-
Partitional: k-means, spectral clustering.
-
Density-Based: DBSCAN (clusters as dense regions).
-
-
Matrix Representation:
-
Adjacency Matrix $A$: $$\displaystyle A_{ij} = 1 $$ if edge $i \to j$, else 0.
-
Incidence Matrix: Node-edge connections.
-
Laplacian Matrix: $$\displaystyle L = D - A $$, where $D$ is degree matrix.
-
2.7 Supporting Technologies & Concepts
-
Email Groups:
Mailing lists for group communication; archived discussions; used for community building.
-
RSS Feeds:
Really Simple Syndication: XML-based format for content distribution (e.g., news, blogs). Users subscribe to feeds for automatic updates.
3.0 Digital Image Processing – Forensic & Analytical Applications
3.1 Image Formation & Perception
-
Image Formation in the Human Eye:
Light → Cornea → Pupil (aperture) → Lens (focus) → Retina (photoreceptors: rods for brightness, cones for color) → Optic nerve → Brain.
-
Brightness Adaptation and Discrimination:
-
Adaptation: Eye adjusts sensitivity to light (pupil dilation/constriction).
-
Discrimination: Ability to distinguish brightness differences; follows Weber's Law: $$\displaystyle \frac{\Delta I}{I} = \text{constant} $$, where $\Delta I$ is just noticeable difference.
-
3.2 Image Quality & Noise
-
Approaches to Estimate Noise Parameters:
-
Local Statistics: Compute variance in flat regions (noise variance).
-
Filtering: Apply median filter; residual = noise estimate.
-
Transform Domain: Analyze high-frequency coefficients (e.g., DCT, wavelet).
-
-
Image Sampling and Quantization:
-
Sampling: Spatial discretization; pixel grid. Nyquist rate: $$\displaystyle f_s \geq 2f_{\max} $$ to avoid aliasing.
-
Quantization: Amplitude discretization; $$\displaystyle L = 2^k $$ levels for $k$ bits; quantization error $\in [- \Delta/2, \Delta/2]$, $$\displaystyle \Delta = \frac{\text{range}}{L} $$.
-
3.3 Frequency Domain Processing
-
2-D Fourier Transform Linearity Proof:
Let $$\displaystyle f(x,y) = a f_1(x,y) + b f_2(x,y) $$.
$$F(u,v) = \iint f(x,y) e^{-j2\pi(ux+vy)} dx dy$$
$$= a \iint f_1(x,y) e^{-j2\pi(ux+vy)} dx dy + b \iint f_2(x,y) e^{-j2\pi(ux+vy)} dx dy$$
$$= a F_1(u,v) + b F_2(u,v).$$
Similarly for discrete case. $\boxed{\text{Linear}}$
-
Homomorphic Filtering:
Concept: Separate illumination ($i$) and reflectance ($r$) components: $$\displaystyle f(x,y) = i(x,y) \cdot r(x,y) $$.
Steps:
-
Log transform: $$\displaystyle \ln f = \ln i + \ln r $$.
-
Fourier transform: $$\displaystyle F(\ln f) = F(\ln i) + F(\ln r) $$.
-
Filter $H(u,v)$ (highpass for reflectance enhancement).
-
Inverse transform, exponentiate: $$\displaystyle \hat{f}(x,y) = e^{\ln \hat{i} + \ln \hat{r}} $$.
Equations:
-
$$\hat{F}(u,v) = H(u,v) \cdot F(\ln f)$$
$$\hat{f}(x,y) = e^{\mathcal{F}^{-1}\{\hat{F}(u,v)\}}$$
-
Image Sharpening:
- Butterworth Highpass Filter (BHPF):
$$H(u,v) = \frac{1}{1 + \left(\frac{D_0}{D(u,v)}\right)^{2n}}$$
where $$\displaystyle D(u,v) = \sqrt{u^2 + v^2} $$, $$\displaystyle D_0 $$ = cutoff, $n$ = order.
- Gaussian Highpass Filter (GHPF):
$$H(u,v) = 1 - e^{-D^2(u,v)/(2D_0^2)}$$
Both attenuate low frequencies, enhance edges.
3.4 Spatial Domain & Color Processing
-
Histogram Processing of Color Images:
Apply to each channel (RGB, HSV) separately or use 3-D histogram.
Operations:
-
Histogram Equalization: $$\displaystyle s_k = T(r_k) = (L-1) \sum_{j=0}^{k} p_r(r_j) $$ for each channel.
-
Histogram Matching: Match to specified histogram.
-
-
Image Point Operations:
Pixel-wise transformation: $$\displaystyle s = T(r) $$, where $r$ = input intensity, $s$ = output.
Examples:
-
Contrast stretching: $$\displaystyle s = \frac{1}{1 + (c/r)^E} $$ (log), or linear piecewise.
-
Thresholding: $$\displaystyle s = 0 $$ if $$\displaystyle r < T $$, else $L-1$.
-
3.5 Image Restoration
-
Minimum Mean Square Error (MMSE) Filtering:
Estimate $\hat{f}$ from degraded image $g$ to minimize $$\displaystyle E[(\hat{f} - f)^2] $$.
Solution: $$\displaystyle \hat{F}(u,v) = \frac{P_{fg}(u,v)}{P_g(u,v)} G(u,v) $$, where $P$ = power spectra.
-
Wiener Filtering:
Optimal MMSE filter for additive noise:
$$H(u,v) = \frac{P_f(u,v)}{P_f(u,v) + P_n(u,v)}$$
where $$\displaystyle P_f $$ = image power spectrum, $$\displaystyle P_n $$ = noise power spectrum.
[!TIP] Wiener filter balances inverse filtering and noise smoothing.
3.6 Image Segmentation
-
Region-Based Segmentation:
Group pixels based on similarity (intensity, texture).
Methods: Thresholding, region growing, split-and-merge, watershed.
-
Motion-Based Segmentation:
Use motion cues (optical flow) to separate moving objects from static background.
Steps: Compute flow vectors → cluster based on velocity.
-
Differentiation:
| Aspect | Region-Oriented | Motion-Based | |---------------------|--------------------------------------|---------------------------------| | Basis | Spatial intensity/color similarity | Temporal motion differences | | Input | Single frame | Sequence of frames | | Use Case | Static scenes | Video, dynamic scenes |
3.7 Morphological Processing
-
Basic Operations (for binary image $A$, structuring element $B$):
-
Erosion: $$\displaystyle A \ominus B = \{ z \mid (B_z) \subseteq A \} $$; shrinks objects.
-
Dilation: $$\displaystyle A \oplus B = \{ z \mid (B_z) \cap A \neq \emptyset \} $$; expands objects.
-
-
Morphological Algorithms:
-
Boundary Extraction: $$\displaystyle \beta(A) = A - (A \ominus B) $$.
-
Hole Filling:
$$\displaystyle X_0 = \text{seed point inside hole} $$;
$$\displaystyle X_{k+1} = (X_k \oplus B) \cap A^c $$;
iterate until $$\displaystyle X_{k+1} = X_k $$;
then $$\displaystyle A \cup X_k $$ fills hole.
-
3.8 Image Compression
-
Need for Compression:
Reduce storage and transmission bandwidth; exploit redundancy (spatial, spectral, psycho-visual).
-
Vector Quantization (VQ) Method:
-
Codebook: Set of prototype vectors (trained on image blocks).
-
Encoding: Find nearest codevector (minimum distance) → send index.
-
Decoding: Replace index with codevector.
Advantage: High compression at fixed bitrate; Disadvantage: Block artifacts.
-
-
Lossy Predictive Coding:
Encoder:
DiagramCANVAS: Block diagram: Input image → Predictor (using past pixels) → Subtracter → Quantizer → Encoder → Output bitstream. Also, Quantized error fed back to predictor.Working: Predict pixel $\hat{f}(x,y)$ from neighbors (e.g., linear prediction); encode prediction error $$\displaystyle e = f - \hat{f} $$; quantize and entropy-code.
Decoder:
DiagramCANVAS: Block diagram: Input bitstream → Decoder → Inverse quantizer → Adder → Reconstructed image. Predictor uses reconstructed pixels to generate prediction.Reconstructs using same predictor and quantized error.
3.9 Image Analysis & Recognition
-
Texture Analysis:
-
Statistical: Gray-Level Co-occurrence Matrix (GLCM) for contrast, homogeneity.
-
Structural: Identify primitives (e.g., edges, spots) and their arrangements.
-
Spectral: Filter-based (e.g., Gabor filters), wavelet transforms.
-
-
Object Recognition:
Steps:
-
Preprocessing: Enhancement, segmentation.
-
Feature Extraction: Shape, texture, SIFT, HOG.
-
Matching/Classification: Template matching, machine learning (SVM, CNN).
Applications: Biometrics, autonomous vehicles, medical diagnosis.
-