Mobility Load Balancing (MLB)

MLB operates by dynamically adjusting the Cell Individual Offset (CIO) between neighboring cells. By modifying this handover trigger threshold, the network artificially expands or contracts the coverage footprint of a cell. A positive CIO offset makes a target cell appear more attractive to User Equipment (UE), encouraging earlier handovers and offloading traffic from a congested source cell. This is a reactive, closed-loop process where the source cell initiates load balancing when its Physical Resource Block (PRB) utilization exceeds a defined threshold.

MLB algorithms rely on specific, monitored metrics to trigger and evaluate actions:

• Load Metric: Typically PRB utilization ratio (uplink and downlink).
• Trigger Threshold: A configurable point (e.g., 70% PRB usage) where the cell is considered overloaded.
• Target Selection: Neighbors are evaluated based on available capacity and current load.
• Handover Success Rate (HOSR): Continuously monitored to ensure load balancing actions don't degrade mobility robustness.
• Call Drop Rate (CDR): A critical safety metric; if CDR increases, MLB actions are rolled back.

A critical design challenge is the inherent conflict between MLB and Mobility Robustness Optimization (MRO). While MLB aggressively shifts handover boundaries to balance load, MRO adjusts the same parameters to prevent Radio Link Failures (RLFs). An MLB-triggered early handover might be interpreted by MRO as a 'too-early handover' event. SON Coordination is required to arbitrate these conflicting goals, often using a weighted cost function or a policy-based priority system where MRO (stability) takes precedence over MLB (capacity) to prevent a cascading network failure.

MLB is standardized in 3GPP TS 36.300 and TS 36.423 for LTE. The procedure relies on X2/Xn interface signaling between base stations (eNBs/gNBs):

• Resource Status Reporting Initiation: A loaded cell requests load reports from neighbors.
• Resource Status Update: Neighbors periodically report their current load levels.
• Mobility Settings Change: The source cell proposes a CIO change to a specific neighbor.
• Mobility Change Acknowledge/Reject: The target cell accepts or rejects the proposed boundary shift based on its own capacity.

Advanced implementations move beyond reactive threshold-based logic to predictive load balancing. By integrating a time-series forecasting model (e.g., an LSTM or Transformer) on the Near-RT RIC, the system can anticipate traffic surges minutes in advance. This allows for proactive CIO adjustments before congestion occurs, smoothing the user experience. The model ingests historical PRB utilization, UE count, and time-of-day features to predict future load states, enabling a seamless transition from reactive to proactive resource management.

MLB is not limited to intra-frequency neighbors. Inter-frequency MLB uses A4/A5 measurement events to redirect UEs to less loaded carriers on different frequencies. Inter-RAT MLB extends this concept to different Radio Access Technologies (e.g., LTE to 5G NR or 3G). This is crucial for heterogeneous networks where capacity layers (e.g., 3.5 GHz) can offload traffic from coverage layers (e.g., 700 MHz). The mechanism relies on dedicated priority-based cell reselection for idle mode and blind redirection or PS handover for connected mode UEs.