How Transit Signal Priority Speeds Up Buses Without New Lanes

Ride a city bus and time the parts. You'll notice something annoying: the bus barely spends any of the trip actually driving. It spends it at stops, at red lights, waiting to merge back into traffic, behind a car turning left.

That's the opening for transit signal priority. If a bus's problem were top speed you'd need a new lane. Its problem is stopping, and some of that stopping happens at signals a city already owns and can reprogram.

TSP is the least glamorous transit improvement there is. Nothing about it photographs. It's also one of the few things an agency can do without acquiring an inch of road.

A bus loses more time standing still than moving

Break a bus trip into four buckets: time actually in motion, dwell at stops while people board, delay at traffic signals, and everything else, which mostly means congestion and the awkward business of pulling out of a stop.

The proportions vary enormously. A suburban route with stops every half mile is dominated by driving time. A downtown route with a stop every 600 feet and a signal at every corner can spend most of its trip stationary.

Component	What drives it	What fixes it
Time in motion	Distance, speed limits, stop spacing	Stop consolidation; little else
Dwell at stops	Boardings, fare payment method, door count, wheelchair ramp cycles	Off-board fare payment, all-door boarding, level boarding
Signal delay	Cycle length, number of signals, stop position relative to the intersection	Transit signal priority, far-side stops, queue jumps
Traffic and re-entry delay	Congestion, curbside conflicts, yield-to-bus rules	Bus lanes, bus bulbs, daylighting, enforcement

That table is a structure, not a measurement. The point is that signal delay is one bucket of four and TSP only touches that bucket, which is the most important thing to understand before anyone promises you a faster bus.

The four things TSP can actually do

Strip away the vendor language and there are four moves.

Green extension

The bus is approaching, the light is green, and it's about to turn yellow. The controller holds the green for a few extra seconds so the bus gets through instead of stopping.

This is the best move available, and it's not close. It's cheap because the green was already running, so nobody on the cross street loses a whole phase. And it converts a full stop into no stop at all: no braking, no waiting out a red, no accelerating a heavy vehicle back to speed.

Early green, also called red truncation

The bus is approaching, the light is red, and the controller cuts the conflicting phase short to bring the bus's green up sooner. The bus still stops, just for less time.

Weaker than extension: you've shortened a delay rather than deleted it, and it's more disruptive, since you're taking green from someone mid-phase. Still worth having.

Phase insertion or rotation

Reordering the phases in a cycle, or inserting one, so the bus's movement comes up next rather than fourth. Useful at complex intersections with protected lefts. Harder to deploy without confusing drivers, who learn an intersection's rhythm and dislike it changing.

Queue jump

A short stretch of bus-only or right-turn-shared lane at the approach, plus a signal giving the bus a few seconds of head start. The bus pulls ahead of the queue and merges in front. Not purely signal work since it needs pavement, but on a corridor where the bus otherwise sits in a long right-turn backup, a queue jump at three bad intersections can outperform TSP at twenty ordinary ones.

Conditional priority is the whole game

Early deployments granted priority to every bus that asked. That's unconditional priority, and it's a mistake for a reason that's obvious once stated: it helps the buses that need it least.

A bus running four minutes early doesn't want to go faster. It'll sit at a timepoint and wait anyway, so every second of priority it consumed came out of the cross street for nothing. Worse, on a frequent route, speeding up an early bus makes bunching worse: the gap behind it grows, and the following bus picks up more passengers, dwells longer, and falls further behind.

Conditional priority asks questions before granting a request:

• Is this bus actually behind schedule, or below the target headway from the bus ahead?
• Does it have passengers aboard, or is it deadheading to the garage? And how many? A crush-loaded bus represents far more person-minutes than a near-empty one.
• Has this intersection already granted priority recently, or is it owed recovery time?
• Is the cross street queued past the point where taking green would cause real harm?

That logic is where the value lives, and it's also where projects fail. Conditional priority needs the bus to know its own schedule adherence and to communicate it, which means the transit agency's vehicle location system has to talk to the traffic department's signal system in real time. Those are two different organizations with two different procurement cycles and, frequently, two different opinions about who's in charge.

If a city tells you it has installed transit signal priority, the follow-up question is whether it's conditional. Unconditional TSP on a frequent route can make reliability worse while making the average trip marginally faster, and reliability is what riders actually notice.

How the signal knows a bus is coming

Four generations of technology, all still in service somewhere.

Inductive loops and transponders. A loop in the pavement reads a tag on the bus. Reliable, dumb, fixed. The bus is either over the loop or it isn't, so the controller gets no advance warning and no idea how far away the bus is.

Optical and infrared emitters. A strobe or IR emitter on the bus, a receiver at the intersection, same family as emergency vehicle preemption. Line-of-sight dependent, so a truck ahead of the bus can block the request, and historically vulnerable to people buying an emitter online and giving themselves greens.

GPS plus automatic vehicle location. The bus knows where it is and how late it is, reports to a central system, and that system asks the controller for priority. This is what makes conditional priority possible. Latency is the enemy: a request arriving three seconds late has missed the window entirely.

Connected vehicle messaging. Signals broadcast their phase and timing, buses broadcast position and intent, and the negotiation happens with far better information. It enables something loops never could: telling the driver what speed to hold to arrive on a green, rather than only telling the signal to change. That's underrated. An advisory to hold 22 mph for the next three greens produces a smoother ride, less fuel burn, and fewer priority requests, which means less impact on cross traffic for the same time saving.

Green extension versus early green

If a city can only implement one, it should be green extension, and the reasoning is worth spelling out because it's pure arithmetic.

A bus that gets a green extension saves the entire remaining red plus its own deceleration and acceleration. On a 90-second cycle, arriving just after green ends might cost 40 to 60 seconds of red, plus 10 to 15 seconds of speed loss and recovery for a loaded bus. Extending green by five seconds can therefore save close to a minute for a cost of five seconds taken from the cross street.

Early green has a much worse ratio. To save a bus 15 seconds of red you take roughly 15 seconds from a conflicting movement. You're moving delay around rather than deleting it, which only pays if the bus carries meaningfully more people than the traffic you're delaying.

One practical consequence: extension only helps buses arriving in a narrow window near the end of green, so its hit rate is low. A corridor with extension only will show modest average improvements. That's expected, not failure. It saves a lot of time for a minority of trips.

What the cross street loses

Priority isn't free and honest engineers don't pretend otherwise. Every second granted comes from somewhere, and the constraints are firmer than people assume.

The hard floor is pedestrian crossing time. A signal must provide enough green for someone to cross at an assumed walking speed, plus a clearance interval, and that minimum can't be shortened to make room for a bus. On a wide cross street there's very little to borrow. On a narrow side street there's more.

Then recovery. After granting priority, a signal is out of step with its neighbors and spends the next cycle or two clawing back into coordination. During recovery the cross street can suffer worse delay than the priority event itself caused, which is why controllers cap how often priority can be granted.

The general finding is that cross-street delay increases stay small at moderate volumes and grow sharply once an approach nears capacity. So TSP works well on corridors with slack and poorly where every approach is saturated. Those saturated corridors are, of course, where the buses are slowest.

Why TSP so often disappoints

Cities install it, publish a press release, and then riders report that the bus feels the same. Usually one of these is why.

Near-side stops

This is the big one. If the bus stop sits before the intersection, the bus stops to pick up passengers, and by the time it's ready to go the green it was granted has expired. You paid for priority and handed it back. Moving stops to the far side of the intersection so the bus clears the signal first and then dwells is often worth more than the signal work itself, and it's a fraction of the cost.

Slow fare payment

A bus that dwells 40 seconds because everyone is paying at the front door with cash cannot be rescued by a five-second green extension. Dwell reduction and signal priority multiply each other; doing only one is a waste of the other.

Schedule padding that eats the savings

Suppose TSP saves 90 seconds across a route. If the schedule wasn't rewritten, the bus arrives at its next timepoint early and holds there for 90 seconds. Total trip time: unchanged. The savings existed and the timetable ate them. A TSP project isn't finished until the schedule is rebuilt around the faster running time.

Priority that was quietly turned off

Engineers who get complaints about cross-street delay sometimes shrink the priority window, tighten the conditions, or disable it at problem intersections. That's a legitimate response to a real problem, but it means the system on paper and the system in operation can diverge substantially, and nobody publishes when a setting changes.

Cheap fixes that beat TSP

If a corridor has a fixed budget, TSP frequently isn't the first thing to buy. Ranked roughly by value per dollar:

1. Stop consolidation. Removing stops 400 feet apart eliminates whole dwell events plus the acceleration and braking around them. Politically painful, mechanically the most effective single change available.
2. Far-side stop relocation. Cheap. Makes TSP work. Reduces the re-entry conflict too.
3. All-door boarding with off-board or tap payment. Attacks the largest bucket on busy urban routes.
4. Bus bulbs. Extend the curb so the bus stops in the travel lane instead of pulling in and merging back. Eliminates re-entry delay.
5. Daylighting and parking removal at stops. Paint and signs, and it removes the maneuvering that costs 10 seconds a stop.
6. Then TSP.
7. Then a bus lane, if the corridor's problem turns out to be congestion rather than any of the above.

So the technology project belongs in the middle of the list, not the top. TSP is popular partly because it doesn't require taking anything visible away from drivers, which makes it easier to approve than the changes that would help more.

The coordination problem nobody mentions

Traffic signals almost always belong to a city or county public works department. Buses belong to a transit agency, which may be regional and may cross a dozen municipal boundaries on one route.

So a corridor-length project can need agreement from several signal owners, each with their own controller hardware, firmware versions, communications backhaul, and view on whether buses deserve priority over local traffic. A single corridor can carry three generations of equipment. That's why TSP timelines look absurd relative to the engineering involved. The engineering is well understood. The interagency agreements and the controller upgrades are where the years go.

Measuring it honestly

TSP is unusually easy to evaluate, because the buses are already reporting where they are and when. The metrics worth demanding, in order of usefulness:

• Running time distribution, not average. The 95th percentile trip matters more to riders than the mean, because that's the one they plan around.
• Excess wait time on frequent routes. How much longer people wait than the headway implies, which captures bunching. If TSP made this worse, priority probably isn't conditional.
• Intersection-level grant rates. How often a request was made and how often it was honored. This is where you discover priority was disabled at the three worst intersections.
• Cross-street delay and queue length at those same intersections, so the cost side is visible.
• Whether the schedule was rewritten. If it wasn't, no rider received any benefit regardless of the running-time data.

Cost, and who has to agree

Costs scale with what's already in the ground. A city with modern controllers, reliable communications to each cabinet, and a central signal system is doing a software and configuration project. A city with 1990s controllers and no communications is doing capital construction that happens to end in some software.

The case that actually persuades boards isn't rider time savings, even though that's the largest benefit. It's operating cost: if a route's round-trip running time drops enough, the agency can hold the same frequency with fewer buses, or raise frequency with the same fleet. Vehicle hours dominate bus operating costs, so a durable cut in cycle time converts directly into savings or more service.

Which is the other reason the schedule rewrite isn't optional. Without it there's no operational saving, no rider benefit, and a lot of equipment doing nothing useful.

If you're trying to get this done in your city, the sequence that works is: pick one corridor, move the near-side stops first, get the AVL data flowing before you buy anything, insist the priority logic is conditional on both lateness and headway, and write into the project scope that the timetable gets rebuilt on completion. Skip that last clause and you'll have a faster bus that arrives at exactly the same time.

Questions

Is this the same as emergency vehicle preemption?

No, and the distinction is important. Preemption interrupts the signal immediately and unconditionally, dropping everything to serve a fire truck, then recovers. Priority nudges an existing cycle within limits, respects pedestrian minimums, and can be refused. Preemption is disruptive by design because the stakes justify it; priority has to be polite because buses come every eight minutes.

Does TSP help streetcars and light rail too?

Yes, and often more, because rail vehicles accelerate slowly, carry more people, and can't route around an obstruction. The same conditional logic applies. Street-running rail with poor signal priority is especially frustrating, since the capital cost was already paid.

Will it make my drive worse?

If you're on the corridor the bus runs along, marginally better or unchanged, since fewer bus stops in your lane means less obstruction. If you're crossing it, slightly worse, concentrated in the cycles where priority was granted. How much depends heavily on how close your approach is to capacity.

Why doesn't every city just do this?

Because it requires the traffic department and the transit agency to share operational control of an asset, and institutional arrangements are harder to change than firmware. The technical barriers were solved a long time ago.