Best Place To BuildHow does 5G work? | A RARE look inside the 5G testbed facility @IITM | BP2B Labcast Ep 1
CHAPTERS
Labcast kickoff at IIT Madras: inside India’s 5G testbed
The host introduces the series and sets the purpose of the visit: to understand what IIT Madras built in its 5G testbed and why it matters. The episode frames the facility as the site of India’s first 5G phone call and promises a practical look at real 5G infrastructure.
- •Labcast Episode 1 premise and goals
- •Visit to IIT Madras 5G testbed facility
- •Why this lab is significant (India’s first 5G call context)
- •What viewers will learn: how 5G works, what’s being built now
5G signal path explained: phone → antenna → baseband → core → internet
A simplified end-to-end explanation is given for how a 5G connection is established. The segment outlines the roles of the radio/antenna, the baseband unit, and the 5G core in authenticating users, managing mobility, and routing data.
- •Phone communicates via radio signals to an antenna/radio unit
- •Signals get processed into digital form for further handling
- •Baseband unit manages the connection request and radio resources
- •5G core authenticates, manages mobility, and routes to internet/other users
Radio unit hardware: why 5G uses many more antenna elements
Jeeva walks through the radio unit and contrasts 4G antenna element counts with 5G’s much larger arrays. The discussion connects more antenna elements to better directionality, capacity, and performance—while also increasing calibration complexity.
- •4G often uses ~1–4 antenna elements; 5G commonly 16/32/64
- •More elements enable better user-focused transmission (beamforming concept)
- •Energy is directed rather than broadcast wastefully in all directions
- •Higher complexity requires precise tuning and calibration per element
Beamforming & calibration: directing energy without disrupting service
The conversation explains why multi-antenna systems must be continuously calibrated and why that’s non-trivial in real deployments. Calibration must adapt to changing conditions (environment, temperature, aging) without interrupting live traffic.
- •Radio doesn’t ‘know’ user direction; multiple antennas allow steering energy
- •Calibration must be algorithmic and minimally invasive to ongoing transmission
- •Frequency of calibration depends on channel dynamics and deployment scenario
- •Environmental factors: temperature variation and hardware aging
What’s inside a tower radio: modular 16 vs 32 antenna systems + thermal design
The team describes typical operator deployments (e.g., 32-antenna setups) and shows how their design scales modularly. They also cover the practical engineering needed outdoors—especially heat dissipation without fans.
- •Typical commercial deployments may use ~32 antenna systems
- •Their 16-antenna unit uses modular internal digital processing blocks
- •Scaling to 32 involves additional aggregation/processing modules
- •Fanless thermal management using heat pipes and dissipation fins
Timing & synchronization: GPS/GNSS and base-station time alignment
This chapter explains why precise timing is critical for cellular networks and how systems synchronize using satellite signals. Sync ensures neighboring base stations coordinate in time to avoid interference and maintain reliable handovers.
- •GNSS/GPS antenna provides timing reference to the base station
- •Time synchronization is required even when systems are frequency-separated
- •Base stations must align timing to reduce interference and coordinate operation
- •Satellite timing acts as a common reference across the network
Why 5G can outperform 4G: massive MIMO and parallel signal chains
The episode links 5G speed/capacity gains to larger MIMO configurations and parallel processing. It clarifies MIMO as multiple simultaneous input/output chains whose combined processing boosts throughput and reliability.
- •4G commonly referenced as 8x8 MIMO in the discussion
- •5G scales to 16x16, 32x32, up to 64x64 MIMO
- •MIMO = multiple parallel transmit/receive chains
- •Signals from many antennas are processed jointly to extract the user data
Baseband unit responsibilities: resource allocation, handover, and low latency
The baseband unit’s job is detailed: allocating frequencies/time slots, controlling power/gain, and managing handovers. The segment positions the BBU as the real-time controller that keeps connections stable as users move.
- •Assigns radio frequencies and time resources
- •Controls power and gain settings
- •Manages tower-to-tower handover for mobility and low latency
- •Splits processing between radio-side and baseband-side components
Open interfaces and splits: O-RAN 7.2 and where processing happens
The team describes how O-RAN defines standardized “splits” so radios and basebands from different vendors can interoperate. Their implementation uses an O-RAN 7.2 split, with some processing on the radio (FPGA) and the rest in the baseband physical layer.
- •O-RAN enables multi-vendor compatibility via standard interfaces
- •Processing is divided by a defined functional split (7.2)
- •Radio unit uses FPGAs for part of signal processing
- •Remaining PHY processing runs in the baseband unit
Fighting noise in real time: channel estimation every 0.5 ms
This chapter gets into the “secret sauce” of making wireless work under motion and noise. The system repeatedly estimates the channel extremely frequently (about every 0.5 milliseconds) to separate desired signals from interference and adapt to changing conditions.
- •Algorithms distinguish user signal (“hello”) from noise/interference
- •Channel estimation is continuous, not occasional
- •Estimation periodicity discussed: ~0.5 milliseconds
- •Enables reliable service while moving (train/rickshaw scenarios)
Inside the 5G core: software-defined network functions at campus scale
The core network is presented as the brain/backbone that manages users, policies, and routing. The episode contrasts older hardware-centric cores with 5G’s software-based core that can run on general-purpose compute, scaling from laptops to carrier-grade deployments.
- •Core manages user information, policies (e.g., recharge), and routing
- •Campus core aggregates traffic across IIT Madras deployments
- •Shift from custom hardware (older generations) to software-based 5G core
- •Scalability: laptop-scale demo to thousands/millions of users in operator cores
How the indigenous 5G testbed came together (2018 consortium → 2022 first call)
The researchers describe the longer academic groundwork and the 2018 government-led push that united IITs into a consortium. They recount the multi-year build culminating in April 2022, when a commercial 5G phone first latched onto their base station and made the official call.
- •Academic research predated 2018; 2018 catalyzed coordinated execution
- •Government convened IITs; consortium formed to build state-of-the-art stack
- •Four-year build to complete the testbed; first official call in April 2022
- •Breakthrough moment: phone successfully connected after parameter tuning
The people and disciplines behind the stack: RF, thermal, hardware, FPGA, simulation
This segment highlights that “end-to-end 5G” requires many specialized teams, not just software. They outline who builds what—from mechanical chassis and thermal design to RF, PCBs, FPGA code, and spec-driven simulation work.
- •Thermal + mechanical teams design chassis and cooling
- •RF team designs antennas and power amplification chain
- •Hardware team builds PCB/electronics platforms
- •FPGA/software teams implement the protocol/codebase
- •Simulation team maps 3GPP specs into MATLAB/C/Python models and validates performance
What they’re improving now: mobility, optimization, and the bridge to 5G-Advanced/6G
The conversation shifts to ongoing development: improving mobility performance and optimizing transmit/receive algorithms. The lab is also applying lessons learned to next-generation radio/baseband designs that can evolve toward 5G-Advanced and 6G with even larger antenna arrays.
- •Current testing focus: high-speed mobility with minimal latency
- •Continuous optimization of receiver and transmitter algorithms
- •New radio/baseband units under development beyond the original testbed
- •Expectation that 6G may require many more MIMO/radio elements
AI/ML in 5G/6G research: dynamic scenarios like centimeter-level positioning
They explain why AI/ML can outperform conventional methods in highly dynamic or hard-to-model situations. A key example is precise user/device positioning (centimeter-level) for industrial indoor use cases, potentially reducing measurement time via learning-based methods.
- •AI/ML helps where environments and use cases are highly dynamic
- •Example use case: centimeter-level positioning from base stations
- •Industrial scenario: tracking products on conveyor belts indoors
- •Goal: reduce measurement time and improve accuracy with AI-based approaches
Looking ahead: reducing reference-signal overhead and scaling to massive IoT
The episode closes with open problems and 6G aspirations. They point to the bandwidth cost of reference signals used for channel estimation and anticipate a future with many low-power IoT nodes connecting alongside traditional user devices.
- •Challenge: heavy reference-signal overhead consumes bandwidth
- •Research direction: reduce known-signal overhead while maintaining performance
- •6G vision includes far more IoT nodes with small processing capabilities
- •Future networks may support massive numbers of connected transceivers
