https://github.com/cn-upb/deepcomp - deepcomp 1.4.2

DeepCoMP is now accepted for publication in the 2023 IEEE Transaction on Network and Service Management (TNSM) as "Multi-Agent Deep Reinforcement Learning for Coordinated Multipoint in Mobile Networks" 🎉

• Updated Readme
• Fix dependencies for correct installation: Pin protobuf and pydantic

- Python
Published by stefanbschneider almost 3 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.4.1

• Enable 'avg' reward aggregation for DeepCoMP by default (was sum)
• Add --debug CLI option for running in a debugger

- Python
Published by stefanbschneider over 4 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.4.0

Improvements regarding utility functions: * Use constants to define the max and min utility, which are then applied for normalization, reward clipping, and rendering * Support two additional utility functions (in addition to log): Linear (ie, just data rate) and step function. * Configurable via CLI. But: Requires manual adjustment of MINUTILITY and MAXUTILITY

Full Changelog: https://github.com/CN-UPB/DeepCoMP/compare/v1.3.0...v1.4.0

- Python
Published by stefanbschneider over 4 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.3.0

• Two, configurable heuristics: Dynamic and static
• Configurable dynamic UE arrival and departure over time
• Changed reward function for multi-agent: Weighted avg. QoE over all cells in range (based on their connected UEs)
• Added observation for multi-agent: Avg. QoE of connected UEs at each cell
• Multiple smaller changes, fixes, eg, upgrade to Ray 1.4

- Python
Published by stefanbschneider almost 5 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.2.5

Updated Docker support: https://hub.docker.com/r/stefanbschneider/deepcomp

- Python
Published by stefanbschneider about 5 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.2.4

• Critical bug fix in CLI
• Improved formatting
• Docker support

- Python
Published by stefanbschneider about 5 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.2.3

Another small fix in the Readme

- Python
Published by stefanbschneider about 5 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.2.2

Minor fix in readme for PyPi

- Python
Published by stefanbschneider about 5 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.2.1

Minor fixes in Readme and rendered video.

- Python
Published by stefanbschneider about 5 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.2.0

• Much improved and extended render function for nicer visualization. New --dashboard mode, new icons, metrics, etcs.
• New, restructured CLI args
• Clean up issues (with DeepSource)
• Minor other changes

- Python
Published by stefanbschneider about 5 years ago

https://github.com/cn-upb/deepcomp - deepcomp 1.1.0

• Update to ray 1.2
• New CLI features, eg, for multi-node cluster, simplified videos, etc.
• Update readme, setup, license

- Python
Published by stefanbschneider about 5 years ago

https://github.com/cn-upb/deepcomp - PyPi Release

Release of deepcomp package on PyPi. Install via

pip install deepcomp

Functionally equivalent to v1.0. Now using semantic versioning for new releases.

- Python
Published by stefanbschneider over 5 years ago

https://github.com/cn-upb/deepcomp - Major release v1.0

Major release of DeepCoMP, DD-CoMP, and D3-CoMP

- Python
Published by stefanbschneider over 5 years ago

https://github.com/cn-upb/deepcomp - Cooperative Multi-Agent

• New observation space with better normalization improving performance of both central and multi agent PPO
• Extra observations and new reward function for multi agent PPO to learn non-greedy, cooperative & fair behavior, taking other UEs into account
• Support for continuous instead of episodic training
• Refactoring, fixes, improvements

Details: v0.10 details

- Python
Published by stefanbschneider over 5 years ago

https://github.com/cn-upb/deepcomp - Preparation for Evaluation

• New variants for observation (components, normalization, ...) and reward (utility function and penalties)
• New larger scenario and adjusted rendering
• New utility scripts for evaluation: Running experiments and visualzing results
• Bug fixes and refactoring
• Default radio model is resource-fair again (more stable than proportional-fair)

Details: v0.9 details

- Python
Published by stefanbschneider almost 6 years ago

https://github.com/cn-upb/deepcomp - Proportional-fair sharing, Heuristic baselines, Improved Env

• Support for proportional-fair sharing (new default)
• 2 new greedy heuristic algorithms as baselines
• New default UE movement: Random waypoint
• New default UE utility: Log function with increasing data rate
• Improved and refactored environment and model

Details: v0.8 details

- Python
Published by stefanbschneider almost 6 years ago

https://github.com/cn-upb/deepcomp - Larger Environment, CLI support

• Larger environment with 3 BS and 4 moving UEs.
• Extra observation (optional) showing number of connected UEs per BS. To help learn balancing connections. Seems not to be very useful.
• Improved visualization
• Improved install. Added CLI support.

Details: v0.7 details

- Python
Published by stefanbschneider almost 6 years ago

https://github.com/cn-upb/deepcomp - Multi-agent RL

• Support for multi-agent RL: Each UE is trained by its own RL agent
• Currently, all agents share the same RL algorithm and NN
• Already with 2 UEs, multi-agent leads to better results more quickly than a central agent

Details: v0.6 details

- Python
Published by stefanbschneider almost 6 years ago

https://github.com/cn-upb/deepcomp - Improved radio model and observations

• Improved radio model: Configurable sharing/fairness models for multiple UEs connected to a BS. New default: Rate-fair sharing.
• Improved observations: Extra observation indicating the current total data rate of each UE combined over all its connections (normalized)
• New penalty for losing connection rather than disconnecting actively
• Many smaller improvements and adjustments

Details: v0.5 details

- Python
Published by stefanbschneider almost 6 years ago

https://github.com/cn-upb/deepcomp - RLlib

• Replaced stable_baselines with ray's RLlib, which is more powerful and supports multi-agent RL
• Major refactoring of most code
• No changes in radio model or MDP

Details: MDP description

- Python
Published by stefanbschneider almost 6 years ago

https://github.com/cn-upb/deepcomp - 2ue-2bs-central-agent-simple-radio

• Multiple moving UEs
• Controlled by single, centralized agent that sees combined observations and takes combined actions for all UEs
• Updated radio model: Split rate among connected UEs, allow connecting from farther away, data rate of connections adds up

Details: MDP description

- Python
Published by stefanbschneider almost 6 years ago

https://github.com/cn-upb/deepcomp - 1ue-2bs-simple-radio

• Simplest case works: Just 1 moving UE, 2+ fixed basestation.
• Simple radio model for calculating SNR based on distance to BS. Then calculate achievable data rate from SNR. Interference supported but currently disabled. No schedules yet.
• Advanced observation space using clipping and normalization (configurable)
• Baseline RL algorithm (PPO) learns to connect the UE to always at least one BS as it moves

Details: MDP description

- Python
Published by stefanbschneider about 6 years ago

https://github.com/cn-upb/deepcomp - 1ue-2bs-no-radio

• Simplest case works: Just 1 moving UE, 2+ fixed basestation.
• No radio/wireless model implemented yet. Instead, each BS just has a fixed radius of coverage.
• Baseline RL algorithm (PPO) learns to connect the UE to always at least one BS as it moves

- Python
Published by stefanbschneider about 6 years ago