Essential Edge Computing Architectures

Why This Matters

Edge computing architectures aren't just technical implementations—they represent fundamentally different approaches to solving the core challenge of modern computing: where should processing happen? You're being tested on your ability to understand latency optimization, resource distribution, network topology, and scalability trade-offs. Each architecture makes different assumptions about connectivity, device capabilities, and workload characteristics, and knowing these distinctions is what separates surface-level memorization from genuine understanding.

When you encounter these architectures on an exam, don't just recall their names. Ask yourself: What problem does this architecture solve? Where does processing occur? How does it handle the tension between local responsiveness and centralized power? The items below are grouped by their fundamental design philosophy—master these categories, and you'll be able to reason through any scenario an FRQ throws at you.

---

Cloud-Extended Architectures

These architectures take traditional cloud capabilities and push them toward the network edge. The core principle: maintain cloud-like services while reducing the physical and logical distance data must travel.

Fog Computing

• Extends cloud to the network edge—creates an intermediate layer between end devices and centralized data centers for localized processing
• Reduces bandwidth consumption by filtering, aggregating, and processing data locally before selective cloud transmission
• Supports heterogeneous deployments across IoT networks, smart city infrastructure, and industrial environments requiring real-time analytics

Mobile Edge Computing (MEC)

• Positions compute at cellular base stations—processing occurs within the radio access network itself, enabling sub-millisecond response times
• Optimized for mobile-first applications like augmented reality, live video streaming, and location-based services requiring context-aware processing
• Standardized by ETSI as a key enabler for 5G networks, making it a frequently tested architecture in telecommunications contexts

Cloudlets

• Small-scale data centers at strategic locations—think of them as mini-clouds deployed in public spaces, retail environments, or enterprise campuses
• Designed for resource-constrained mobile devices that need to offload intensive computation while maintaining low latency
• Supports VM-based workloads with rapid instantiation, enabling dynamic service deployment closer to end users

---

Network-Integrated Architectures

These architectures leverage existing network infrastructure to embed computing capabilities. The key insight: networks aren't just pipes for data—they can be platforms for processing.

Multi-access Edge Computing

• Spans multiple network types simultaneously—integrates cellular, Wi-Fi, and fixed-line connectivity into a unified edge computing platform
• Enables seamless handoffs between access technologies while maintaining consistent application performance and service continuity
• Critical for smart transportation and autonomous systems where devices frequently transition between network environments

Edge-Cloud Hybrid Architecture

• Dynamic workload orchestration—intelligently distributes tasks between edge nodes and cloud resources based on latency requirements, computational complexity, and cost
• Balances the trade-off between edge responsiveness and cloud scalability through policy-driven resource allocation
• Essential for burst workloads where edge handles steady-state processing but cloud absorbs demand spikes

---

Decentralized Architectures

These architectures distribute control and resources across multiple independent nodes. The fundamental principle: resilience and scalability through elimination of single points of failure.

Peer-to-Peer Edge Computing

• No central coordination required—edge devices share resources directly, creating a self-organizing mesh of computational capability
• Maximizes resilience by distributing workloads across peers; if one node fails, others absorb its tasks without service interruption
• Ideal for collaborative applications like distributed machine learning, content delivery networks, and federated data analysis

Distributed Edge Computing

• Geographically dispersed processing nodes—computing resources spread across multiple physical locations to minimize data travel distance
• Optimizes for reliability through redundancy; critical for applications like smart grid management and industrial monitoring where downtime is unacceptable
• Supports predictive maintenance by processing sensor data locally and only escalating anomalies to central systems

---

Tiered and Specialized Architectures

These architectures organize edge resources into structured layers or optimize for specific use cases. The design philosophy: match processing location to task requirements.

Hierarchical Edge Computing

• Multi-tiered resource organization—simple tasks handled at lower tiers (closest to devices), complex analytics escalated to higher tiers with more capability
• Optimizes resource allocation by routing workloads based on computational complexity, urgency, and data sensitivity
• Reduces unnecessary data movement by filtering and aggregating at each tier before passing to the next level

IoT Edge Computing

• Purpose-built for sensor networks—processes high-volume, low-complexity data streams from connected devices at the network periphery
• Dramatically reduces cloud bandwidth costs by performing local filtering, compression, and preliminary analysis before transmission
• Enables immediate actuation for time-critical applications like industrial automation, smart home systems, and safety-critical monitoring

Edge-Centric Computing

• Edge-first design philosophy—treats edge nodes as primary processing locations rather than supplements to cloud infrastructure
• Prioritizes local intelligence for applications requiring guaranteed low latency and high availability, such as autonomous vehicles and real-time medical devices
• Minimizes cloud dependency by keeping sensitive data and critical processing at the edge whenever possible

---

Quick Reference Table

---

Self-Check Questions

1. Which two architectures both extend cloud capabilities to the edge but differ in whether resources are concentrated (discrete locations) or distributed (mesh across nodes)?

2. If an autonomous vehicle needs guaranteed sub-millisecond response times while transitioning between cellular and Wi-Fi networks, which architecture best addresses this requirement, and why?

3. Compare and contrast Peer-to-Peer Edge Computing and Hierarchical Edge Computing in terms of their approach to workload distribution and fault tolerance.

4. An industrial facility needs to process thousands of sensor readings per second, filter out normal readings locally, and only send anomalies to a central system. Which architecture category best fits this use case, and what specific architecture would you recommend?

5. FRQ-style prompt: Explain how Edge-Cloud Hybrid Architecture balances the trade-offs between latency and scalability. Provide a specific application scenario where this architecture would outperform a purely edge-centric or purely cloud-centric approach.