By László Fisi
Field Applications Engineer, Future Electronics (Hungary)
LED light sources are highly efficient and offer the potential to dramatically reduce energy consumption and operating costs compared to conventional incandescent lamps. But while the replacement of conventional lamps with LED lamps dramatically reduces the amount of electricity that a lighting installation uses, there is more that lighting users and specifiers can do
to save energy.
One of the main advantages of LED lighting is that it is very easy to control the light output of LEDs linearly and instantly, from nearly 0% up to maximum brightness. So in fact, the scope for energy savings is even greater if an intelligent control scheme is applied to a lighting installation. For instance, LED street lights can be made to react to the intensity of ambient light (sunlight or moonlight), supplementing the natural light to reach a desired lux value in the illuminated area. Use of proximity sensing allows the lights to be dimmed when no-one is in or near the illuminated area.
Today’s LED drivers support various means for adjusting the duty cycle of an LED lamp, controlled either by a microcontroller’s PWM signal, or a 0-10V DC or 1-10V DC output. There is therefore high potential for intelligent control of street lights, but to realise this potential, the designer of the lighting system must provide a networked communication scheme through which to exercise remote control of the lights.
Wide-area control network for lighting
It would, in theory, be possible to install completely autonomous, unconnected lighting units, each with its own ambient light sensor (photometer), proximity sensor and control circuit. But this architecture has two disadvantages:
• every lighting unit carries the cost of the sensing components
• central monitoring, control and diagnostic functions are not supported
So it is, in fact, more desirable to connect street lights to a control network. Not only does this enable the operator to detect instantly when a lamp fails, so that it can be repaired quickly to avoid prolonged downtime; it also allows for central control of an entire lighting scheme. For instance, a municipal authority might wish to override the normal ambient/proximity light sensor controls at special times, such as firework displays, festivals or big cultural or sporting events.
In the past, exterior lighting networks used a dedicated network cable (such as an RS-485 link) for connectivity, but the cost of supplying and installing additional signal cabling alongside the power cabling makes this option unattractive. This led to efforts to implement Power Line Communications (PLC), in which data signals are carried on the same copper cable as the power. The reliability of such PLC systems, however, has been called into question, because communication is prone to failure or disruption in the presence of noise and interference on the power line.
By contrast, wireless RF communication eliminates the cost and time involved in installing data cabling, and with sophisticated channel-hopping and other techniques is highly resistant to interference.
The challenge in a street lighting network is to implement RF communications successfully over a wide area, in which the total distance from the central control point to the farthest node could be tens of kilometres. For reasons of cost, the RF transmissions should be carried in licence-free ISM bandwidth. Mobile telephone or cellular networks offer guaranteed wide-area coverage, but the cost of equipment and monthly data plans effectively rules them out for cost-sensitive lighting applications.
But ISM-band radios typically have a maximum point-to-point span of 2km. Wide-area coverage therefore requires a communications protocol which can support multi-path routing of data packets, using intermediate nodes to forward packets to nodes on the edge of the network, while avoiding transmission failures caused by collisions between packets.
A successful street-lighting installation, supported by Future Electronics in Hungary, provides a model for the use of ISM bandwidth to connect hundreds of lighting units over a wide area to a lighting control unit. In this installation, photo-sensors were placed on a small number of units, and their readings were used to control the dimming of lighting units nearby, which could be assumed to be exposed to similar levels of ambient light.
Network protocol avoids collisions
Each lighting unit contains a SPIRIT1 RF transceiver supplied by STMicroelectronics, which can be configured to operate at various frequencies below 1GHz. In boosted power mode, 16dBm maximum Transmit power, the SPIRIT1 IC is capable of transmitting over a maximum range of 3km in open space. This transmission range varies depending on the bit rate chosen by the user and on the user’s Transmit power configuration.
Figure 1 (above) outlines the hardware design of each street light node. As it shows, the power supply is configured so as to provide a 0-10V DC or 1-10V DC analogue signal controlled by a microcontroller’s PWM channel. This analogue voltage controls the dimming of the LED light source.
When implementing a multi-path RF network, the communications protocol has to offer the following features:
• conflict resolution
• support for packet repetition
• route mapping and route optimisation
• ‘island mode’ operation in case of communication breakdown
• error reporting
In particular, a solution had to be found for two common problems found in multi-path wireless networks: the hidden-node problem and the exposed-node problem. In this type of complex network, packets are transmitted along a chain of nodes, each node in turn passing the message on to a neighbouring node. Unlike a star network, for instance, this topology has no central hub or router controlling the timing and routing of every transmission.
In the case of a hidden node, A cannot see C; both A and C might both attempt to transmit to B at the same time, leading to a conflict and a failed transmission, as shown in Figure 2. In the exposed node problem, A requests permission to transmit a message to B. When B clears A to send its message, C thinks that the channel is busy, and so delays sending to D a message that it wishes to transmit. In fact, the channel between D and C is not busy.
One of the advantages of the SPIRIT1 transceiver is that its packet structure supports the use of advanced anti-collision network protocols which eliminate the hidden node problem, as shown in Figure 3.
The preamble performs a synchronisation function, while the credit specifies the packet’s life so that packets cannot circulate indefinitely. The priority is indicated by the third block. The source and destination blocks contain the source and destination addresses. The control block contains the control byte packet descriptors. The load contains the actual information to be transmitted; and the checksum is what it says it is.
In fact, the SPIRIT1 transceiver can process in hardware certain other packet structures, such as those carried by Wireless Metering Bus (WMBus) networks and ST’s own STack protocol. However, both the WMBus and STack protocols have 1-byte address fields, limiting a network to a maximum of 256 nodes. For a large-scale street lighting network, a 2-byte address field is required.
For this reason, the installation in Hungary adopted the MACAW (Multiple Access with Collision Avoidance for Wireless) protocol, which is supported by the SPIRIT1 transceiver. A MACAW transmission follows this process:
• The sender asks whether it can send a packet to the receiver, by transmitting an RTS (Request To Send) frame. This makes all nodes around the sender remain silent temporarily in order to avoid a collision.
• The RTS frame contains the address of the receiver. The receiver gets it, and sends back a Clear To Send (CTS) frame, which means that the receiver is ready for the packet.
• The sender sends the data packet to the receiver
• The receiver answers with an ACK or NACK frame to acknowledge receipt or to tell the sender that the frame has not been received.
Both the hidden-node and exposed-node problems are eliminated by the application of the MACAW protocol. A hidden node might not receive the sender’s RTS frame, but it will receive the receiver’s CTS frame, and then remain silent to keep the channel between sender and receiver clear.
In the case of an exposed node, it is possible for A to send an RTS to B at the same time as D sends an RTS to C, causing a collision, as shown in Figure 2. In this case, the MACAW protocol ensures that B and C are both aware of the collision. After an interval, the length of which is determined by a random number generator, the senders begin negotiating for access to the channel again.
Intelligent routing for fast data transmission
The other important feature of a MACAW network is its provision for intelligent routing. All nodes are within range of one or more other nodes. Whenever a new node joins the network, it sends a discovery message to neighbouring nodes, which forward it on until it reaches the network access point. In the case of this street lighting application, a wireless gateway at the lighting operator’s control centre.
Each time the message is forwarded, the credit score is reduced by one, and the value of the received signal strength and routing data are added to the packet, as shown in Figure 4. The credit score starts at 256, so theoretically a linear end-to-end range of 256km could be achieved, for instance, with a string of 256 nodes in which each node is 1km away from its neighbour.
Multiple discovery packets then arrive at the access point. It selects the route to the new node that has the best combination of features, such as highest credit score, adequate signal strength at all nodes. This path will be used in future for all communications between the access point and the new node.
In practice, this intelligent routing scheme results in better network performance than is achieved in full-blown mesh networks, in which all data packets are always transmitted to all nodes. Mesh networks are commonly swamped by traffic, leading to repeated collisions and re-transmissions. In a complex mesh network, it can take as long as 30 minutes for a packet to travel from one edge of the network to another.
The network path management supported by the MACAW protocol loads the network with much less traffic, and transmission times can be reliably kept down to a fraction of a second even across networks with hundreds of nodes and an edge-to-edge span of more than 100km. Should a transmission path fail, the receiver node simply sends a new discovery message in order to configure a new route.
The protocol also provides a fail-safe mechanism: if communication fails permanently, the microcontroller in the lighting unit switches to ‘island mode’. In this mode, it operates autonomously, switching the lamp on and off in real time according to a pre-programmed pattern.
Central control, superior energy savings
The RF network implemented for the street lighting installation in Hungary provides for real-time control and monitoring of every lighting unit’s status, energy consumption and so on. It also enables enhanced energy savings (and therefore cost savings) by dimming the lamps in response to inputs from ambient light sensors or proximity sensors. Because of the network, inputs from one sensor can be used to control multiple lighting units.
The installation’s designers achieved a remarkable combination of high network performance and low cost.
• The SPIRIT1 transceiver is a low-cost device that uses licence-free ISM bandwidth, but that is extremely well supported with sofware libraries, drivers and other development resources provided by STMicroelectronics.
• By using RF technology, the installers avoided the cost of installing network cabling.
• The MACAW protocol implemented in this installation provides for quick and reliable data transmission, giving the lighting system operator instant visibility and control of every lighting unit over a wide area, while avoiding collisions.