EMMI reduces edge communication for multimodal AI by 32 times
EMMI: Edge Multi-Modal Intelligence for Communication-Efficient MLLM Inference via Fused Representation Compression
Machine LearningDistributed, Parallel, and Cluster Computing
Summary
Running large AI models that understand images, text, and sensor data on small devices like smartphones or cameras is very hard because these models need a lot of computer power and memory. The paper presents EMMI, a way to shrink and combine data from different sensors right on the device, so only a small summary is sent to a bigger server for understanding. This method cuts the data sent over the network by 32 times and still keeps the AI’s accuracy similar, making it much faster to get answers under low network conditions. It also keeps private data safer by not sending raw sensor information.
What this means in practice
- •For mobile device developers: Send compressed, fused sensor data from phones to cloud AI models for faster, low-bandwidth multimodal reasoning.$Commercial implications: Enables new or improved cloud-assisted multimodal AI apps on mobile devices with constrained bandwidth and power.
- •For industrial iot engineers: Reduce communication overhead by sending compact fused sensor representations from edge sensors to central servers for real-time multimodal analysis.
Authors
Motahare Mounesan, Irfan Khan
Abstract
Recent advances in multimodal large language mod- els (MLLMs) have opened new opportunities for edge intelligence by enabling reasoning across heterogeneous sensor modalities, such as vision, text, and telemetry data. However, deploying these capabilities on resource-constrained edge platforms remains challenging due to the substantial computational, memory, and communication demands of modern MLLMs. Rather than transmitting raw sensor observations or partitioning neural networks at intermediate layers, Edge Multi-Modal Intelligence (EMMI) communicates a compact representation between edge devices and server resources, enabling communication-efficient edge MLLM inference. To achieve this, EMMI performs modality-specific encoding, cross-modal representation fusion, and learned compression at the edge, transmitting only a compact latent representation to server-side resources for high-capacity MLLM reasoning. This representation-centric design reduces communication overhead, preserves local data privacy, and provides a fixed-size interface between heterogeneous edge devices and server-side MLLMs. Evaluation on a representative multimodal benchmark demonstrates that EMMI can reduce the communication payload by 32x while maintaining comparable downstream accuracy, resulting in up to a 3.4x reduction in estimated end-to-end inference latency under bandwidth-constrained edge conditions.