From co-located systems to true integration
Early research often treated sensing and communication as co-located but separate functions that shared a site or an antenna tower. Integrated Sensing and Communication (ISAC) goes further: the same time-frequency-spatial resources, the same radio-frequency chains and the same baseband pipelines serve both purposes under a joint performance objective. The architectural consequence is that resource management, beam management, waveform design and higher-layer signaling must all become sensing-aware.
A practical 6G ISAC architecture therefore spans the radio unit, the distributed unit, the central unit and the service exposure layer. Sensing measurements are generated close to the antenna, refined as they move up the stack, and exposed to applications through controlled interfaces that respect privacy, security and operator policy.
Mono-static, bi-static and multi-static modes
Architecture choices begin with geometry. In mono-static mode a single node transmits and receives echoes. This mode is attractive for coverage continuity and simple deployment but demands high transmit-receive isolation and careful self-interference management. Bi-static mode separates the illuminator from the receiver, reducing isolation pressure and enabling favorable geometries for certain targets. Multi-static or network sensing coordinates multiple transmitters and receivers so that a target is illuminated and observed from several angles; fusion then improves detection probability, localization accuracy and resilience to blockage.
6G networks already maintain tight synchronization for coordinated multi-point and carrier aggregation. That existing synchronization fabric becomes an enabler for multi-static ISAC, provided phase stability and absolute time accuracy meet sensing requirements.
Functional split and processing placement
High-resolution range-Doppler processing and direction-of-arrival estimation are computationally intensive. Architectures must decide which steps run in the radio unit (low latency, high bandwidth to the DU), which run in the distributed unit, and which are deferred to edge or central cloud resources. Latency-sensitive safety applications favor local processing; complex multi-static fusion and long-term scene mapping can tolerate higher latency and benefit from larger computational pools.
Open interfaces such as those defined in O-RAN discussions need extensions or profiles that carry sensing-specific measurements, quality indicators and configuration parameters without overloading the fronthaul. Compression of sensing data cubes and selective reporting of detections rather than raw samples become important efficiency mechanisms.
Control and user plane implications
The control plane must schedule sensing resources alongside communication resources. This includes allocating time-frequency blocks for sensing-centric symbols or pilots, configuring beam patterns that illuminate regions of interest, and negotiating sensing quality of service with applications. The user plane continues to carry data traffic; sensing results may ride on the same plane as measurement reports or may use a dedicated sensing data path toward application servers.
Mobility and handover procedures also change. A sensing track that spans multiple cells requires continuity of identity and state across radio units. Architecture designs therefore include sensing context transfer and track hand-off logic analogous to communication mobility management.
Security, privacy and exposure
Continuous environmental sensing raises privacy concerns. Architectural safeguards include data minimization at the source, purpose limitation through policy engines, differential privacy or aggregation before exposure, and strong authentication and authorization on sensing APIs. Security threats include spoofing of sensing returns, jamming of dual-function waveforms and side-channel leakage of user location through sensing reports. Defense-in-depth—physical-layer authentication, encrypted measurement transport and anomaly detection on sensing streams—belongs in the reference architecture from the start.
A well-specified 6G ISAC architecture treats sensing as a native network capability rather than an afterthought bolted onto a pure communication system. That treatment simplifies standardization, procurement and multi-vendor interoperability while giving operators clear control over the performance and privacy trade-offs that sensing introduces.
