Digital Communications

Coordinated Scheduling

Q: How does coordinated scheduling differ from joint transmission in CoMP?

In coordinated scheduling and beamforming (CS/CB), user data is available at and transmitted from only the serving cell, while the scheduling and beamforming decisions are coordinated across the cooperating set to avoid creating interference toward other cells' edge users. In joint transmission (JT), the same data payload is shared across multiple transmission points over the backhaul and sent simultaneously to the user, providing a coherent or non-coherent combining gain. CS/CB only needs channel state information exchange, so its backhaul demand is on the order of a few hundred kilobits per coordinated user per subframe, whereas JT must move the full user data to every cooperating point, demanding tens of megabits per second per user and tight phase-level synchronization. CS/CB tolerates X2 backhaul latencies up to a few milliseconds, while coherent JT typically requires sub-microsecond timing alignment and a latency budget under about 1 ms.

Q: What backhaul latency can coordinated scheduling tolerate before the gain disappears?

Coordinated scheduling relies on channel state information that ages between the moment it is reported and the moment the coordinated decision is applied. For a user moving at 3 km/h at 2.6 GHz, the Doppler shift is about 7 Hz and the channel coherence time is roughly 50 to 60 ms, so an X2 round-trip latency of 5 to 10 ms still leaves the CSI useful and CS/CB delivers most of its gain. At 30 km/h the coherence time falls to about 2 to 3 ms, and a 10 ms backhaul delay makes the CSI stale, collapsing the cell-edge gain from the typical 20 to 30 percent down to near zero. This is why practical CS/CB deployments target sub-2 ms ideal or near-ideal backhaul and restrict coordination to low-mobility cell-edge users.

Q: How much cell-edge throughput gain does coordinated scheduling actually provide?

Field trials and 3GPP evaluations of CS/CB report 5th-percentile (cell-edge) user throughput gains of about 15 to 30 percent over an uncoordinated reuse-1 baseline, with mean cell throughput gains of roughly 3 to 10 percent. The gain is largest in interference-limited dense deployments with a low number of coordinated cells (typically a set of 3) and degrades as the cooperating set grows because CSI feedback overhead and coordination complexity rise faster than the interference benefit. Coordinated scheduling improves the worst users by trading a small amount of resource at the strong-signal users, so the cell-center throughput change is usually within a few percent.

/koh-OR-dih-nay-tid SKEJ-ool-ing/

As a downlink mode of coordinated multipoint (CoMP), this technique lets a cluster of neighboring cells share channel state information so that a scheduler jointly decides which user each cell serves on every time-frequency resource. Unlike joint transmission, the payload still leaves only the serving cell; coordination simply steers beams and resource assignments so that the interference each cell radiates lands on its own non-scheduled resources rather than on a neighbor's cell-edge user. Paired with coordinated beamforming as the CS/CB CoMP scheme, it delivers roughly 15 to 30% cell-edge throughput gain while needing only a few hundred kilobits of backhaul exchange per coordinated user per subframe.

Category: Digital Communications

CoMP Scheme: CS/CB

Cell-Edge Gain: 15 to 30%

How Coordinated Scheduling Suppresses Intercell Interference

Coordinated scheduling emerged in 3GPP LTE-Advanced Release 11 as one of four CoMP categories alongside joint transmission, dynamic point selection, and coordinated beamforming. The driving problem is intercell interference at the cell edge of a frequency-reuse-1 network: a user near the boundary between two cells receives a wanted signal that is barely stronger than the interfering signal from the adjacent cell, so its signal-to-interference-plus-noise ratio (SINR) can sit at or below 0 dB. Rather than reusing frequency more conservatively, which wastes spectrum, coordinated scheduling keeps reuse-1 but lets the cells agree, on a per-subframe basis, which physical resource blocks each one will use for its edge users so that the strongest interferers stay quiet on those resources.

The mechanism depends on accurate, fresh channel state information. Each user feeds back a precoding matrix indicator, rank indicator, and channel quality indicator not only for its serving cell but for the dominant interfering cells in its measurement set. The coordinating scheduler, which may be centralized at a baseband hotel or distributed with exchanges over the X2 interface, uses these reports to compute a resource assignment that maximizes a network utility such as proportional fairness across the whole cluster. Because no user data crosses the backhaul, the link budget for coordination is modest, but the decisions are only as good as the CSI is current, which makes backhaul latency and user mobility the dominant constraints.

In practice operators restrict the cooperating set to two or three cells and to low-mobility users, since the coordination gain falls off sharply once the reported channel ages beyond the coherence time. The technique pairs naturally with coordinated beamforming, where each cell additionally shapes its transmit nulls toward the victim users identified by the shared CSI, and together they form the CS/CB scheme that dominates real deployments because it avoids the stringent synchronization that coherent joint transmission demands.

Governing Relationships

Cell-edge SINR with coordination:
SINR = P_s|h_s|² / (N₀ + ∑_i∈C β_i P_i|h_i|²)

Coordination mutes interferer i → β_i ≈ 0 on the victim's resource block, so the denominator drops toward N₀.

Proportional-fair scheduling metric:
k* = arg max_k [ R_k(t) / T_k(t) ]

Usable-CSI condition (mobility):
τ_backhaul < T_c ≈ 0.423 / f_d, f_d = v × f_c / c

Where P_s, P_i = serving and interfering power; h = channel gain; β_i = coordination muting factor; C = cooperating set; R_k = instantaneous rate; T_k = average rate; T_c = coherence time; f_d = Doppler shift; v = speed; f_c = carrier. Example: v = 3 km/h, f_c = 2.6 GHz → f_d ≈ 7.2 Hz, T_c ≈ 59 ms, so a 5 ms X2 delay is comfortably usable.

CoMP Scheme Comparison

CoMP Scheme	Data Sent From	Backhaul Need	Latency Tolerance	Sync Requirement	Typical Cell-Edge Gain
Coordinated Scheduling / Beamforming (CS/CB)	Serving cell only	CSI only (~0.1 to 0.5 Mbit/user/subframe)	Up to a few ms	Frame-level	15 to 30%
Dynamic Point Selection (DPS)	Best point, switched per subframe	Data at candidate points + CSI	~1 to 4 ms	Subframe-level	10 to 25%
Non-coherent Joint Transmission	Multiple points, power combined	Full data to all points (tens of Mbit/s)	< 1 to 2 ms	Symbol-level	20 to 40%
Coherent Joint Transmission	Multiple points, phase-aligned	Full data + precoders	< 1 ms	Sub-microsecond phase	30 to 60%
Reuse-1 baseline (no CoMP)	Serving cell only	None	n/a	n/a	0% (reference)

Common Questions

Frequently Asked Questions

How does coordinated scheduling differ from joint transmission in CoMP?

In CS/CB the user data is transmitted from only the serving cell while scheduling and beamforming are coordinated across the cluster to avoid interfering with neighbors' edge users. Joint transmission shares the full payload across multiple points over the backhaul and sends it simultaneously. CS/CB needs only CSI exchange (~0.1 to 0.5 Mbit per coordinated user per subframe) and tolerates a few ms of X2 latency, whereas coherent JT moves tens of Mbit/s per user and demands sub-microsecond phase alignment.

What backhaul latency can coordinated scheduling tolerate before the gain disappears?

The gain hinges on CSI staying fresh relative to the channel coherence time. A 3 km/h user at 2.6 GHz sees T_c ≈ 50 to 60 ms, so a 5 to 10 ms X2 round trip preserves most of the gain. At 30 km/h, T_c falls to about 2 to 3 ms and a 10 ms delay makes the CSI stale, collapsing the cell-edge benefit toward zero. Practical CS/CB therefore targets sub-2 ms near-ideal backhaul and low-mobility edge users.

How much cell-edge throughput gain does coordinated scheduling actually provide?

3GPP evaluations and field trials report 5th-percentile user throughput gains of roughly 15 to 30% over a reuse-1 baseline, with mean cell throughput up only about 3 to 10%. The gain is largest for a tight cooperating set of three cells and shrinks as the set grows because CSI overhead rises faster than the interference benefit. Because it helps the worst users by reallocating resource, cell-center throughput typically changes by only a few percent.

Related Terms

← Coordinated Multipoint Coordinated Turn →