WO2012042454A1

WO2012042454A1 - Device and method for reducing delay of data packet transmissions in wireless networks

Info

Publication number: WO2012042454A1
Application number: PCT/IB2011/054203
Authority: WO
Inventors: Javier Espina Perez; Daniel Martin Gorgen; Tim Corneel Wilhelmus Schenk; Oscar Garcia Morchon
Original assignee: Koninklijke Philips Electronics NV
Current assignee: Koninklijke Philips NV
Priority date: 2010-10-01
Filing date: 2011-09-23
Publication date: 2012-04-05
Anticipated expiration: 2013-04-01
Also published as: TW201218695A

Abstract

For reducing an end-to-end delay of data packet transmissions in a large-scale wireless mesh network, a device, a system and a method are provided for controlling data packet transmissions in the wireless network, wherein an answertime-out of a sender node is adjusted based on a distance between the sender node and a destination node.

Description

DEVICE AND METHOD FOR REDUCING DELAY OF DATA PACKET TRANSMISSIONS IN WIRELESS NETWORKS

FIELD OF THE INVENTION

The invention relates to a device, a system and a method for controlling data packet transmissions in a wireless network.

BACKGROUND OF THE INVENTION

Recently, wireless mesh networks attract more and more attention, e.g. for remote control of illumination systems, building automation, monitoring applications, sensor systems and medical applications. In particular, a remote management of outdoor luminaires, so-called telemanagement, becomes increasingly important. On the one hand, this is driven by environmental concerns, since remote control systems or so-called telemanagement systems enable the use of different dimming patterns, for instance as a function of time, weather conditions and season, allowing a more energy-efficient use of the outdoor lighting system. On the other hand, this is also driven by economical reasons, since the increased energy efficiency also reduces operational costs. Moreover, the system can remotely monitor power usage and detect lamp failures, which allows for determining the best time for repairing luminaires or replacing lamps.

Current radio-frequency (RF) based wireless solutions use either a star network topology or a mesh network topology. In a star network, a data collector has a direct communication path to every node in the network. However, this typically requires a high- power/high-sensitivity base-station-like controller with high-rise placement (e.g. on top of a building), which makes the solution cumbersome to deploy and expensive. In a mesh network, the plurality of nodes does in general not communicate directly with the controller, but via so-called multi-hop communications. In a multi-hop communication, a data packet is transmitted from a sender node to a destination node via one or more intermediate nodes. Nodes act as routers to transmit data packets from neighboring nodes to nodes that are too far away to reach in a single hop, resulting in a network that can span larger distances. By breaking long distances in a series of shorter hops, signal strength is sustained. Consequently, routing is performed by all nodes of a mesh network, deciding to which neighboring node the data packet is to be sent. Hence, a mesh network is a very robust and stable network with high connectivity and thus high redundancy and reliability.

In the prior art, mesh network transmission techniques can be divided in two groups: flooding-based and routing-based mesh networks. In a flooding-based mesh network, all data packets are forwarded by all nodes in the network. Therefore, a node does not have to make complicated routing decisions, but just broadcasts the data packet. By these means, the technique is quite robust. However, in large networks, the data overhead due to forwarding impacts the overall data rate. Moreover, collisions of data packets are more likely to occur, further reducing the overall performance. Hence, the main problem of this solution is the scalability. Routing-based mesh networks can be further divided into proactive and reactive schemes. In proactive routing-based mesh networks, all needed network paths are stored in routing tables in each node. The routing tables are kept up to date, e.g. by sending regular beacon messages to neighboring nodes to discover efficient routing paths. Although the data transmission is very efficient in such kind of network, the scalability is still low, since in big networks, the proactive update of the routing tables consumes large parts of network resources. Moreover, the routing tables will grow with the scale of the network. In addition, the setup of the network requires time and resources in order to build up the routing tables. Reactive schemes, in contrast, avoid the permanent overhead and large routing tables by discovering routes on demand. They use flooding to discover network paths and cache active routes or nodes. When routes are only used scarcely for single data packets, flooding the data packets instead of performing a route discovery might be more efficient. If routes are kept long enough to avoid frequent routing, reactive schemes degenerate to proactive schemes. An example for a reactive routing-based mesh network is used in ZigBee. However, the main problem of this protocol scheme is still the scalability of the network.

In large-scale multi-hop networks, the number of hops a data packet has to travel is large as compared to a hop distance in small networks. In a large radio frequency telemanagement system comprising thousands of nodes, 20-40 hops are likely to occur. However, the delivery chance of an individual data packet decreases with its hop distance, since with every hop, there is a chance that the data packet gets lost.

Hence, a big disadvantage in common wireless mesh networks is constituted by the very limited network scalability. This is due to the fact that every data packet or message is transmitted multiple times due to the forwarding, whereby the overall network throughput is reduced. Also, data packet collisions are more likely to occur causing data packet losses, further reducing the overall performance. Thus, improving the success and reliability of multi-hop end-to-end transmissions is particularly crucial in large-scale multi- hop networks, such as street illumination systems with a high number of luminaire nodes, since end-to-end retransmissions are far more resource/bandwidth costly and delay intensive than in typical smaller networks. Hence, efficient routing protocols and reduction of end-to- end delays are required for large-scale wireless mesh networks in order to achieve the required throughput, response times and robustness.

In order to determine whether a data packet is lost, data packet transmissions are commonly performed in acknowledgement mode. In a hop-by-hop acknowledgement mode, every hop of the multi-hop transmission is confirmed by the receiving node to the preceding transmitting node. However, this leads to high network load. Thus, often end-to- end acknowledgements are used, wherein the final destination node confirms the receipt of the data packet to the initial sender node. In this mode, the sender node waits for a predetermined time, so-called acknowledgement time-out, before retransmitting the data packet for which it was expecting the acknowledgement. In general, the acknowledgement time-out is fixed and common to all nodes of the network. Since the acknowledgement timeout for data packets travelling a short distance is then the same as for data packets travelling a long distance, the delay of the retransmission is unnecessarily increased for short-travelling data packets, affecting the overall transmission speed of the network. This causes large delays as well as delay differences in the communication between any two nodes in the network, leading to a poor user experience due to the high and/or heterogeneous delays, e.g. when interacting with the luminaire nodes of an illumination system. If this disadvantage were addressed by simply reducing the size of the network, scalability would sink further. Therefore, an end-to-end delay of successful data packet delivery should be minimized.

US 2009/0154395 Al describes a routing method for a clustered mesh network for minimizing transmission delays by sharing performance measures with neighboring nodes. By these means, the next node is selected, to which the data packet is to be forwarded in the multi-hop transmission.

SUMMARY OF THE INVENTION

In view of above disadvantages and problems in the prior art, it is object of the present invention to provide a device, a system and a method for transmitting data packets in a wireless network that minimize an end-to-end retransmission delay, while maintaining or even increasing network scalability.

The object is solved by the features of the independent claims. The present invention is based on the idea to minimize an answer time-out for a given pair of sender node and destination node based on a distance between the sender node and the destination node that a data packet has to cover. The answer time-out relates to a waiting time that a sender node waits for an answer to a data packet transmitted to a destination node. If the sender node has not received an answer data packet after this time period has passed, the sender node will retransmit the data packet. This eliminates

unnecessary communication delay between retransmissions of unacknowledged data packets.

In one aspect of the present invention, a device for a node of a wireless network is provided for controlling data packet transmission, when the node operates as a sender node. The device comprises a control unit that can adjust the answer time-out based on a distance between the sender node and the destination node. The answer time-out defines a time interval, during which a sender node waits for an answer to a previously sent data packet. When the answer time-out has passed without the sender node having received an answer from the destination node, the sender node will start retransmission of the data packet. By individually adjusting the answer time-out for each pair of sender node and destination node, the delay of detecting a failed transmission is reduced, thus decreasing the end-to-end delay of a successful transmission due to retransmissions close to the minimum possible value. Thus, in a large-scale lighting system, this will decrease the delay of control commands, so that luminaire nodes will react faster, e.g. to dimming or switching commands.

In a preferred embodiment, the answer time-out relates to an

acknowledgement time-out, i.e. the time interval a sender node waits for an

acknowledgement of the destination node indicating successful data packet transmission, before retransmitting the data packet. Possibly, the acknowledgement from the destination node contains additional data, next to the acknowledgement part. By these means, the networking is made more efficient, since no further packet needs to be sent containing the additional data. Thus, the acknowledgement time-out may also refer more general to a time interval that a sender node waits for data from the destination node.

Preferably, data packets are transmitted from a sender node to a destination node via intermediate nodes using multi-hop transmissions in end-to-end acknowledgement mode. That means that a successful transmission is acknowledged after the data packet has reached its destination node.

In one embodiment, the device can be added or coupled to an existing node or a control center of the wireless network. Thus, the device is associated with a network node, which may also be a data collector node. The data collector node may be any node that is configured to communicate with a control center of the network and may function as a kind of gateway. For instance, the device may be adapted to be inserted in an existing circuit board or to be connected to an existing controller of the node. This is in particular useful for improving or upgrading an existing system such as a street lighting system. In addition to the control unit, the device may further comprise a memory and/or a transceiving unit for receiving and transmitting data packets.

The wireless network may have a mesh topology, wherein each node may act as a router. Such a network has increased redundancy and reliability. Preferably, the nodes of the wireless network are stationary, as it is mainly the case for large outdoor lighting systems. Alternatively or additionally, the positions of at least some nodes may be known to at least some of the other nodes of the network and/or to a control center of the network. For instance, at least some of the nodes may store a routing table for data packet transmission from the respective node to a closest data collector node. Preferably, a routing protocol for data packet transmission to the closest data collector node is based on many-to-one routing. Hence, a data packet is transmitted to the neighboring node that is closer to one of the data collector nodes. By these means, data packet transmission becomes faster and more efficient. Moreover, this also allows to dispense with multiple data collector nodes in large wireless mesh networks, e.g. in a street lighting system with a number of luminaire nodes exceeding 1000, thereby increasing redundancy and improving the reliability of the network.

In a further embodiment, the distance between two nodes, e.g. the sender node and the destination node or an intermediate node, is defined by a hop distance, a GPS-based distance and/or an Euclidean distance. The hop distance between two nodes may be characterized by the hop count, i.e. the number of hops required for transmitting a data packet between the two nodes, or by the number of intermediate nodes forwarding the data packet to the final destination node. A GPS-based distance may be derived from the GPS positions of the sender node and the destination node, whereas Euclidean distance refers to the spatial distance between the two nodes. The metric of the distance may be chosen according to a routing protocol applied in the network. If the routing protocol uses a hop count metric, it will be easy to determine the hop distance between two nodes. Likewise, when the network addresses of the nodes are related to their geographic or GPS position, it will be

advantageous to use a GPS-based or Euclidian distance to define the distance between the sender node and the receiver node.

Preferably, the communication functions of the control unit can be sub-divided into different layers. Then, a lower protocol layer, e.g. the medium access control (MAC) layer, may be adapted to determine whether the data packet transmission remains

unacknowledged. A higher protocol layer of the control unit, e.g. a network layer, transport layer or application layer, may be adapted to consider information available in an underlying lower protocol layer, or the other way around. For instance, the network layer may use parameters determined by the MAC layer. By means of this cross-layer communication, the system becomes more reliable and flexible. In one embodiment, the acknowledgement timeout is adjusted by a higher protocol layer that is responsible for end-to-end retransmission of lost data packets. Furthermore, the distance information related to the distance between the sender node and the destination node may be provided from a lower protocol layer to a higher protocol layer by cross-layer communication. Alternatively or additionally, the distance information may be obtained from a routing table, a hop-counter and/or a time-to-live counter. Possibly, the routing table is stored in the device. The hop-counter or the time-to- live counter may be included in a data packet. Then, a destination node may determine the distance information from the hop counter and/or from the time-to-live counter, in order to adjust an answer time-out for a data packet to be transmitted in the reverse way, i.e. from the destination node to the sender node. However, a hop count or a time-to-live count may also be stored in the sender node for a plurality of destination nodes. The distance information may also be determined using a difference between an initial time-to-live count and a final time-to-live count of a corresponding reverse link. This information may then be stored for future data packet transmissions. Possibly, the initial time-to-live count is known or equal for all nodes of the system. Additionally or alternatively, the distance information may be derived from information received from the destination node and/or from techniques for building-up routing tables for the wireless network.

In a further embodiment, the answer time-out is based on the hop count and a hop time. For instance, the answer time-out can be calculated as: time out =

2 *hop_count*hop_time* constant. Here, the constant may be larger than 1. The hop count can be the hop count of the last data packet received from the destination node, the average of the last n data packets received from the destination node, the maximum hop count over the last n data packets received from the destination node, a moving average of hop counts of the data packets received from the destination node over time, or the like. Likewise, the hop time may correspond to an average hop time required on average for forwarding a data packet from one node to another, i.e. for one hop. Alternatively, the hop time may refer to an average hop time over the last n data packets from the destination node, to a maximum hop time, to a moving average of hop times over time or to a success hop time, within which a predetermined percentage of hops is successfully performed. Moreover, the answer time-out may be adjusted based on a type of the data packet to be sent, e.g. whether it is a time-critical or a time-uncritical data packet or what priority rank the data packet has. For this, the control unit of the device may further be able to determine the type of the data packet. Additionally or alternatively, the answer time-out may be based on the current traffic load in the network, which will result in a higher hop time. This can be determined e.g. by the amount of network traffic the sender node is observing or receiving, or how many data packets the sender node sent out in a recent time period T. Alternatively, the receiver node and/or any of the intermediate nodes can report information about a current network load to the sender node, e.g. as part of a data packet.

In a further preferred embodiment, when a node acts as an intermediate node in a multi-hop data packet transmission, the control unit of the device can adjust transmission parameters for forwarding a received data packet based on a distance that the data packet has already traveled. This increases the probability for a long-traveled data packet to survive the final hops before arriving at its destination node. For instance, the transmission parameters may include a maximum number of retransmissions at a lower protocol layer, a maximum number of medium access attempts, a transmit power level, a delay time for retransmission and/or a back-off time for medium access attempts. Here, medium access attempt relates to the process of carrier sensing and the subsequent transmitting or retransmitting of a data packet, when the medium is assessed to be free. Hence, the back-off time for medium access attempts denotes the time interval between subsequent medium access attempts. Likewise, the delay time for retransmission refers to the time between subsequent retransmissions. The transmit power level is related to the signal strength of the transmitted data packet. Thus, long-traveled data packets will be prioritized on the expense of short-traveled data packets, resulting in a reduced end-to-end-delay for data packet transmissions between distant nodes and in an inherent homogenization of the end-to-end delay in the network. Especially in large-scale luminaire networks, this will have the advantage of synchronized luminaire behavior, e.g. in response to a broadcast dimming command.

Data packet transmission may be performed by wireless radio -frequency (RF) transmissions. Since RF transmissions do not require high transmission power and are easy to implement and deploy, costs for setting up and operating a network using the device can be reduced. This is especially important for large networks, e.g. a telemanagement network for lighting systems. However, data packet transmission may alternatively use infrared communication, free-space- visible-light communication or powerline communication. In a preferred embodiment, the device is used in luminaire nodes of an outdoor lighting system for telemanagement of luminaire nodes. Thus, the luminaire nodes can be easily switched on/off and/or the diming pattern of the luminaire nodes can be controlled based on parameters, such as daytime, season, weather, ambience brightness, occurrence of traffic accidents, presence of road works, etc. Possibly, at least some of these parameters are determined by sensors provided with the luminaire nodes and reported to a control center.

In another aspect of the present invention, a system is provided for controlling data packet transmissions in a wireless network. The system comprises a control center and a plurality of nodes. At least one of the control center and the nodes comprises the device according to one of the above-described embodiments. The control center is adapted to control the function or operation of the nodes in the wireless network. For instance, when the nodes are associated with the luminaires of a lighting system, e.g. a street lighting system, the control center may control the nodes individually and/or in groups based on their spatial distribution with respect to their dimming pattern and operation state. Preferably, at least one of the nodes comprises a memory and/or a sensor. If the node comprises a sensor, the node may be adapted to transmit the sensor data to the control center.

In another aspect of the present invention, a method is provided for controlling data packet transmissions in a wireless mesh network having a plurality of nodes. According to this method, a data packet is sent from a sender node to a destination node, wherein the answer time-out for the data packet is adjusted based on a distance between the sender node and the destination node. The order of the steps is arbitrary and can be changed. Hence, the step of adjusting the acknowledgement time-out may be performed before transmitting the data packet. Preferably, this method is applied in a telemanagement system for lighting systems.

BRIEF DESCRIPTION OF THE DRAWINGS

In the drawings:

Fig. 1 shows an example of a wireless mesh network;

Fig. 2A shows a schematic view of the wireless mesh network illustrating hop- distances between nodes;

Fig. 2B shows a schematic view of the wireless mesh network illustrating Euclidean distances between nodes; Fig. 2C illustrates a spatial distribution of nodes in an street illumination

system;

Fig. 3 shows a schematic view of a device according to the present invention; Fig. 4 shows a flow diagram illustrating an embodiment of the present

invention; and

Fig. 5 shows a flow diagram illustrating a further embodiment of the present invention.

DETAILED DESCRIPTION

Preferred applications of the present invention are outdoor lighting systems (e.g. for streets, parking and public areas), indoor lighting systems for general area lighting (e.g. for malls, arenas, parking, stations, tunnels etc.) or sensor networks. In the following, the present invention will be explained further using the example of an outdoor lighting system for street illumination. In the field of lighting control, the telemanagement of outdoor luminaires via radio-frequency network technologies is receiving increasing interest, in particular solutions with applicability for large-scale installations, e.g. with segments of above 200 luminaires.

In fig. 1, a typical network with mesh topology is shown. A plurality of nodes 10 (N) is connected to each other by wireless communication paths 40. Some of the nodes 10 function as data collector nodes 50 (N/DC), which receive data packets from the surrounding nodes 10 via single-hop or multi-hop transmissions and transmit them to a control center 60 and vice versa. Thus, the data collector nodes 50 may operate in the manner of gateways between the nodes 10 and the control center 60. The wireless communication path 40 between the nodes 10 and data collector nodes 50 may be constituted by radio frequency transmissions, while the connection 70 between the data collector nodes 50 and the control center 60 may make use of the Internet, mobile communication networks, radio systems, ethernet, DSL, cable or other wired or wireless data transmission systems.

In a telemanagement system for outdoor lighting control, communication is very asymmetric. Most of the traffic is generated by the nodes 10, e.g. reporting their state, sensor values or power usage to the control center 60. The other traffic consists of control commands from the control center 60 to the different nodes 10, e.g. for adjusting a dimming pattern or switching on/off lamps. Therefore, most traffic is constituted by N-to-1 traffic (unicasts), whereas the traffic from the control center 60 to the nodes 10 consists of 1-to-N traffic, either in unicast, multicast or broadcast mode. Moreover, the number of luminaire nodes 10 is extremely high in an outdoor lighting system such as a street lighting system. Hence, the size of the network is very large, especially when compared to common wireless mesh networks, which typically contain less than 200 nodes. In addition, the nodes 10 have limited processing capabilities due to cost considerations, so that processing and memory resources in the luminaire nodes 10 will be limited. Thus, communication protocols for transmitting data packets between single nodes 10 should consider the limited resources for efficient and fast data packet transmission. Furthermore, compared to other so-called ad-hoc mesh networks, the telemanagement system for an outdoor lighting control network is stationary, i.e. the nodes 10 do not move. Also, all luminaire nodes 10 may be connected to mains power. Consequently, network changes will be mainly due to a changing environment, e.g. due to traffic. Since the nodes 10 are stationary, the physical positions of the nodes 10, for instance GPS coordinates, may be known in the system, enabling geographic or position- based routing. Furthermore, telemanagement of an outdoor lighting system does not require a high data rate. However, there are some scenarios, where a low response time is needed for a certain type of messages or data packets. For instance, when a traffic accident is detected, nodes 10 of the corresponding area can be controlled as to immediately switch to full power.

Due to the specific application properties of an outdoor lighting system as mentioned above, the following features can be applied. The data packet transmission from a data collector node 50 to the respective luminaire nodes 10 can be performed by flooding, wherein all data packets are forwarded by all receiving nodes 10 in the network. The data packet contains at least information about the sender node 10 and one or more destination nodes 10. The data packet is then decoded by the at least one destination node 10. For data packet transmission from the luminaire nodes 10 to the data collector nodes 50, a routing- based solution is preferred, wherein every node 10 selects as intermediate node 10 a neighboring node 10 that is closer to one of the data collector nodes 50. Preferably, a proactive routing structure is used, since the routes to the data collector nodes 50 are regularly used. In the proactive routing structure, a routing table is stored in every node 10, indicating which neighboring node 10 is closer to one of the data collector nodes 50. Thus, data packets can be sent to the closest data collector node 50 in a very efficient and fast way. Advantageously, each node 10 keeps information about multiple downlink neighboring nodes 10 as alternative routes in order to increase reliability. If one neighboring node 10 is not reachable due to strong interference or complete failure, then the routing protocol has additional alternatives to route the data packet to the data collector node 50. In fig. 2A, a data collector node 50 surrounded by a plurality of nodes 10 is shown, illustrating a multi-hop unicast data transmission from a sender node A to the data collector node 50 (destination node B) via a plurality of intermediate nodes Nl ...Ni. The nodes 10 have different hop distances to the data collector node 50 as indicated by radius 501 and 502. For instance, a node A within radius 501, but outside radius 502 will need two hops hi and h2 for transmitting data packets to the data collector node 50 being the destination node B, i.e. a data packet has to be transmitted from this node A to the data collector node 50 via one appropriate intermediate node Nl . In contrast, a node 10 within radius 502 can transmit its data packets in one hop to the data collector node 50. Of course, the destination node B can be any node 10 and is not necessarily a data collector node 50. Thus, a hop distance can be defined for every pair of a sender node A and a destination node B. A parameter for characterizing the hop distance is the hop count, i.e. the number of hops required to transmit the data packet from the sender node A to the destination node B.

In fig. 2B, a Euclidean distance d between the sender node A and the destination node B is illustrated. Between any two nodes 10, the Euclidean distance is defined as the geometric distance between two points. If the network addresses of the nodes 10 are based on the GPS position of the respective nodes 10, also a GPS-based distance may be used. The distance between two nodes 10 is then defined as the distance between their GPS positions. In particular, when the nodes 10 of the network are equally distributed over the network area, a Euclidean or GPS-based distance between two nodes will be

characteristic for the number of hops performed on average when transmitting a data packet between two nodes and thus also for the transmission time. Alternatively, the distance can refer to the actual distance traveled by the data packet. In outdoor lighting networks, a distance measured along streets can be used, rather than a Euclidian distance, since data packets will likely travel along these paths. This is illustrated in fig. 2C, showing street luminaire nodes 10 arranged along streets. Thus, the distance between two luminaire nodes 10 can also refer to a street distance, which is defined as the spatial distance or the hop distance along streets of a road system.

In fig. 3, a device 100 according to the present invention is shown. The device 100 can be associated with a node 10 or data collector node 50 of a wireless multi-hop mesh network, e.g. to luminaires of a lighting system. The device 100 comprises a control unit 200. Moreover, either the node 10 or 50 or the device 100 comprises a transceiving unit 300 for transmitting or receiving data packets via wireless communication paths 40, e.g. via radio frequency transmission. The control unit 200 of the device 100 may be sub-divided into different layers according to its functions in data packet transmission. For instance, when using an OSI-layer model, the control unit 200 will comprise a physical layer for defining the interaction of the device 100 with a transmission medium, a MAC layer providing addressing and channel access control mechanisms in a multi-node network, a network layer providing a plurality of functions and procedures, e.g. network routing functions, a transport layer providing reliable data transfer services to higher protocol layers using e.g. flow control,

segmentation/desegmentation or error control and an application layer for identifying communication partners, determining resource availability or synchronizing communication.

In the following the present invention will be explained using the example of adjusting an acknowledgement time-out, i.e. the time period, for which any sender node waits for an acknowledgment for the sent data packet from the destination node before

retransmitting the data packet. However, the same can be applied for adjusting a general answer time-out, during which a sender node A waits for an answer data packet from a destination node B. Thus, the present invention relates to adjusting an answer time-out and is not limited to acknowledgement time-outs.

In fig. 4, a flow diagram is shown illustrating the present invention using the example of adjusting an acknowledgement time-out, without being limited thereto. When processing a data packet to be sent or when sending a data packet (S40), the distance between the sender node A and the destination node B is determined (S41). Based on this distance, the acknowledgement time-out for this data packet is adjusted (S42). Therefore, when adjusting the acknowledgement time-out, the transport/application layer retransmission in case of a missing acknowledgement is scheduled accordingly. In general, the acknowledgement timeout should take a value as small as possible in order to minimize the end-to-end

communication delay. In the prior art, the acknowledgement time-out is fixed and common for all nodes 10. Hence, the communication delay for closely neighbored nodes 10 is equal to the one for distant nodes 10. However, when adjusting the acknowledgement time-out for each individual sender-destination-combination, as suggested by the present invention, the acknowledgement time-out can be minimized, thus minimizing the end-to-end

communication delay. The minimum value of the acknowledgement time-out equals the expected roundtrip time for a data packet transmission between the source node and the destination node, i.e. the time for delivery of the data packet plus the time for delivery of the acknowledgement. When the sender node A does not receive an acknowledgement for the sent data packet within the acknowledgement time-out, the sender node A will retransmit the data packet (S43). The order of the steps may be changed, i.e. steps S41 and/or S42 can be performed before sending the data packet (S40).

Some of the steps can be performed several times or iteratively in loops, until a maximum number of repeats is reached. For instance, the sender node A can retry the transmission of the data packet (S43) several times, until a maximum number of

retransmissions is reached. If several transmission attempts are allowed and the time-out determined in S42 (determined for the first transmission) is reached, the time-out can be increased for the actual retransmission, e.g. using the following formula:

time out = cl * distance + c2 * n retransmissions, where cl and c2 are two constants and n retransmissions is a retransmission counter, which is 0 for the first transmission when the data packet is transmitted for the first time. This prevents that the time-out is (repeatedly) set too low, which prevents successful transmission.

As mentioned above, the same process is applicable for adjusting any answer time-out. Thus, after a sender node A has transmitted a data packet to a destination node B (S40), the distance between both nodes A and B is determined (S41) and the answer time-out is adjusted accordingly (S42). Again, the order of steps S40, S41 and S42 can be altered arbitrarily. If the sender node A has not received any answer from the destination node B after the answer time-out has passed, it retransmits the data packet (S43). The answer data packet may include an acknowledgement, data, in particular data requested by the sender node A and the like. Also, the acknowledgement from the destination node B can contain additional data next to the acknowledgement part. For instance, when a data collector (sender node A) is sending a request to a receiver node B for sending back the currently used dimming profile. The receiver node B can then send two packets back, i.e. first an

acknowledgement that the request was received and then a data packet containing the response with the current dimming profile. However, advantageously, the acknowledgement and the data about the current dimming profile are combined into one data packet. By these means, no additional transmission is required, thus saving network resources.

The distance between two nodes can be defined using a metric, such as a hop distance, a GPS-based distance or a Euclidean distance. In case that the positions of the nodes 10 are known or that the network addresses are based on GPS positions of the respective nodes 10, the distance information for a particular sender-destination combination can be derived therefrom. Together with an average transmission speed, the GPS-based or the Euclidean distance can be used to estimate an expected value for the roundtrip time, within which an acknowledgement for the sent data packet will be received in case of successful transmission. This value will then determine the acknowledgement time-out. Alternatively, the distance can be derived based on a hop distance between the sender node A and the destination node B. The hop distance is characterized by the number of hops a data packet has to travel to reach its destination node B. In some cases, the hop distance information is already available at a network layer, e.g. when using routing tables with hop count metric. In this case, destination nodes B together with the respective distances from the sender node A are stored in the sender node A.

The distance information can also be explicitly generated, e.g. by the network layer, using a hop counter or a time-to-live counter included in a data packet. The hop counter can be included in a data packet and is increased every hop during the multi-hop transmission from the sender node A to the destination node B. For instance, the sender node A may derive the hop distance information, i.e. a hop count, for a specific destination node B from a hop counter included in a data packet received from the destination node B. Here, the hop count may be derived from the last data packet received from the destination node B or the hop count may relate to the mean or average hop count over the last n data packets received from the destination node B. Alternatively, the hop count may be chosen as the maximum hop count of the last n data packets received from node B or as a sliding-window- average of hop counts of the last n data packets over time. Instead of being included in data packets received from the destination node B, the hop counter can also be included in the acknowledgement from the destination node B for a packet sent by the sender node A, or it can be included in both. Alternatively, the receiver node B can derive the hop count or a derivative of it, such as the average or the mean, from a hop counter of a data packet received from the sender node A and communicate this hop count back to the sender node A. This may be required, when the hop count for one path is expected to be different from the one of the reverse path, i.e. hop count from A to B different than hop count from B to A. The hop count information can be stored at the network layer and used as the hop distance for data packet transmission to this destination node B. Adjusting the acknowledgement time-out is not only applicable to unicast data packets sent to only one destination node B. The embodiment can also be applied in acknowledged broadcast and multicast cases, wherein a data packet is transmitted to several destination nodes B. In these cases, different time-outs for different desitmation nodes B in a multicast group can be taken into account, or the maximum time-out for the nodes in the group is selected.

Similarly, a time-to-live counter can be used. In general, the time-to-live counter (TTL) is a header field with an initial value that is greater than the maximum required number of hops. At each intermediate node Ni, i.e. after each hop, the time-to-live counter is decreased. Data packets with a current time-to-live counter of zero will be dropped in order to avoid infinite forwarding of undeliverable data packets. Thus, a hop count can be derived from the difference between the initial time-to-live count (before the first hop) and the final time-to-live count when receiving the data packet. Here, the destination node B either knows the initial value of the time-to-live counter or the initial time-to-live count is embedded in the data packet. As described above, the final time-to-live count can be extracted from the data packet by the destination node B. Then, the final time-to-live count can be embedded in the acknowledgement for the data packet or in another data packet transmitted the reverse way, i.e. from the destination node B to the sender node A, so that the sender node A can calculate the hop count for the next data packet with the same destination node B.

Of course, distance information and in particular, hop count information can also be generated by the use of other techniques for building-up routing tables, e.g. by regularly sending beacon messages to potential or important destination nodes. Therefore, the adjustment of the acknowledgement time-out can be applicable to dynamic routing protocols, wherein the number of hops to a certain destination node B can vary. The distance information is provided to an upper layer, which deals with end-to-end data packet retransmission, e.g. an application layer or a transport layer. This upper layer can then minimize the acknowledgement time-out value for every destination node or sender- destination-combination. By these means, the end-to-end delay, when a data packet transmission has failed, can be decreased close to the minimum possible value.

When choosing a hop count metric, the acknowledgement time-out can be calculated as: time out = 2*constant*hop_count*hop_time, with the constant being greater than 1. The factor 2 is for the round trip, since the hop count only relates to one way.

Alternatively, a hop count for the round trip may be used instead. Hence, the

acknowledgement time-out is greater than the average roundtrip time. The hop time can relate to an average time period required on average for forwarding a data packet to the next intermediate node Ni in a multi-hop transmission. This may be characteristic for the network. Alternatively, the hop time can depend on the type of the data packet, e.g. whether the data packet is a time-critical or time-uncritical data packet or whether the data packet is labeled with high or low priority. Of course, the hop time can be replaced with other characteristic times, for instance a medium hop time or a success hop time, i.e. a time interval, in which a certain percentage, e.g. 90% -99% of the hops are successfully taken. When using a GPS- based or Euclidean distance, the acknowledgement time-out is similarly calculated using a characteristic time for the chosen metric. It is also possible to take into account some delay in the destination node B, e.g. a buffer time the response data packet is buffered there, a turnaround time from the destination node B to the sender node A, or the like.

Preferably, the acknowledgement time-out can additionally be based on a current traffic load in the network, which will result in a higher hop time. This can be determined e.g. by an amount of network traffic observed or received by the sender node A, or by how many packets the sender node A sent out in the last time period T. Alternatively, the receiver node B (or any of the intermediate nodes Ni) can report information about a network load to the sender node A, e.g. as part of a data packet or together with an acknowledgement.

When applied in a large lighting system with a plurality of luminaire nodes 10, the delay for operation commands can thus be minimized. In particular, for luminaire nodes 10 that are far away from a data collector node 50 (or a control center 60) sending the operation command and that have long transmission times already, this eliminates unnecessary communication delay between retransmissions of unacknowledged data packets. Thus, the scalability of the lighting system can be increased and response times of the luminaire nodes 10 can be decreased.

A further reason for an increased end-to-end delay is caused, when a long- traveled data packet is dropped. Any data packet that has been travelling long across the network has already consumed a significant amount of network resources and accumulated a significant delay. If such a data packet is dropped, it will have to be retransmitted by its sender node A (end-to-end retransmission), which at least doubles the accumulated end-to- end delay and the spending of network resources, i.e. bandwidth. In a multi-hop network, any node 10 may act as a sender node A or as an intermediate Ni forwarding a received data packet to the next intermediate node Ni or to the final destination node B. Therefore, according to a further embodiment of the present invention, a node 10 of the wireless network is not only able to minimize an answer time-out of a data packet based on the distance to the destination node B, when acting as a sender node A. But the node 10 can also adjust transmission parameters for processing a data packet to be forwarded, when the node 10 operates as an intermediate node Ni.

In fig. 5, a flow diagram is shown illustrating the adjustment of transmission parameters for a data packet to be forwarded. In step S50, a data packet is received by the intermediate node Ni, either from the sender node A or from another intermediate node Ni. After having received the data packet, the distance between the receiving intermediate node Ni and the sender node A is determined (S51). Based on this distance, the transmission parameters for the data packet are adjusted. The transmission parameters can refer to MAC parameters, e.g. a maximum number of MAC-layer retransmissions, a maximum number of channel access attempts, a transmit power level, a delay time for retransmission or a back-off time interval for channel access, or a combination thereof. The maximum number of MAC- layer retransmissions determines how often the MAC layer is allowed to retry the

transmission of a data packet. Likewise, the maximum number of channel access attempts relates to the maximum allowed number of times that a MAC layer is allowed to perform carrier sensing in order to get channel access for transmitting a data packet. The delay time for retransmission denotes a delay between subsequent transmission attempts of a data packet and the back-off time interval for channel access refers to a time interval between two subsequent channel access attempts. Thus, the transmission parameters can be adjusted such, that a probability for a long-traveled data packet to successfully pass the final hops to its destination is increased. For instance, the maximum number of MAC layer retransmissions or the maximum number of channel access attempts can be increased or the delay time for retransmission or the back-off time interval for channel access can be reduced. Of course, also a combination of these adjustments can be chosen.

An intermediate node Ni identifies a distance traveled by a data packet on the basis of a hop counter or of a sender address included in the data packet. In the latter case, the network address of the sender node A can be GPS-based. Alternatively or additionally, a routing table comprising the GPS coordinates or the location of the sender node A in the network may be stored in the intermediate node Ni. Long-traveled data packets exhibit high hop count values or a large difference between the GPS positions of the sender node A indicated in the sender address field and the intermediate node Ni, which is at least known locally to the respective intermediate node Ni.

In one example, it is determined that the distance that a data packet has traveled exceeds a certain threshold. Then, a set of preferential transmission parameters for the hop of the data package to the next node 10 can be chosen. Alternatively, the adjustment of the transmission parameters may be proportional to the distance travelled by the data packet. Thus, long-travelled data packets are more likely than shorter traveled data packets to successfully hop to the next node 10. This may even occur at the expense of short-travelled data packets. After having adjusted the transmission parameters based on the distance between the intermediate node Ni and the sender node A, the data packet is processed using these transmission parameters (S53). This may for instance influence an order of a data packet queue, a priority rank parameter, etc., so that a long-traveled data packet may be processed faster. Then, the data packet is forwarded to the next node 10 (S54) using the determined transmission parameters.

By these means, the survival chance of long-travelled data packets is increased. Moreover, data packets perceive comparable delays irrespective of the required number of hops, which is beneficial for an application and also for a transport layer. This is in particular useful, if data packets have to be retransmitted at higher layers in case of data packet loss. Moreover, the answer time-out and in particular the acknowledgement time-out can be set equal and even lower for all data packets. Thus, when transmission parameters are adjusted based on a traveled distance of a data packet, this can be additionally considered for the adjustment of the answer time-out. By these means, the success rate for long-traveled data packets that have already used a lot of network bandwidth can be increased and the total communication delay for long routes can be decreased, possibly at the expense of shorter travelled data packets. Consequently, there is an inherent homogenization of the end-to-end communication delay.

When applied in the telemanagement of an outdoor lighting system, this achieves an increased synchronization of the luminaire nodes 10, since the luminaire nodes 10 will receive instructions with comparable delays. Moreover, the mean and the maximum communication delay for luminaire nodes 10 that are far away from any data collector node 50 (or the control center 60) is decreased, resulting in a higher communication delay homogeneity for luminaire nodes 10 in the same network.

Hence, according to the present invention, the answer time-out can be adjusted for each individual pair of sender node A and destination node B, reducing the end-to-end transmission delay in a wireless network. In addition, when the node 10 acts as an

intermediate node Ni forwarding a received data packet, the intermediate node Ni can adjust the transmission parameters for the data packet based on the distance travelled by this data packet. This helps to decrease the end-to-end delay further and also results in higher delay homogeneity in the network. Thus, overall network resources can be saved.

Claims

CLAIMS:

1. A device for controlling data packet transmissions in a wireless network having a plurality of nodes (10, 50), comprising:

a control unit (200) that is adapted to adjust an answer time-out of a data packet in a sender node (A) based on a distance between the sender node (A) and a destination node (B).

2. The device according to claim 1, wherein an answer data packet, for which the sender node (A) waits during the answer time-out, includes at least one of an

acknowledgement and data.

3. The device according to claims 1 or 2, wherein the answer time-out refers to an end-to-end acknowledgement and/or wherein adjusting the answer time-out is performed by a higher protocol layer responsible for end-to-end retransmissions.

4. The device according to any one of the preceding claims, wherein the device (100) is adapted to be coupled to a node (10) and/or to a data collector node (50) and/or to a control center (60) and/or wherein the wireless network is a mesh network and/or wherein nodes (10) of the wireless network are stationary and/or positions of nodes (10) of the wireless network are known and/or wherein a data packet transmission from the sender node (A) to the destination node (B) is performed in a multi-hop mode via a plurality of intermediate nodes (Ni).

5. The device according to any one of the preceding claims, wherein the distance between two nodes (10, 50) is defined by a hop distance, a GPS-based and/or Euclidean distance.

6. The device according to any one of the preceding claims, wherein a distance information is provided from a lower protocol layer to a higher protocol layer by cross-layer communication and/or obtained from a routing table, from a hop counter and/or a time-to-live counter.

7. The device according to any one of the preceding claims, wherein a distance information is generated by using a hop count of a corresponding reverse link, a difference between an initial time-to-live count and a final time-to-live count of a corresponding reverse link, an information received from the destination node (B) and/or techniques for building-up metric routing tables.

8. The device according to any one of the preceding claims, wherein the answer time-out is based on a hop count and a hop time and/or on a type of the data packet.

9. The device according to any one of the preceding claims, wherein the hop count corresponds to a hop count of a last data packet received from the destination node (B), an average hop count over the last n data packets received from the destination node (B), a maximum hop count over the last n data packets received from the destination node (B) and/or a sliding window average of hop counts received from the destination node (B) over time.

10. The device according to any one of the preceding claims, wherein the hop time corresponds to an average hop time, an average hop time over the last n data packets received from the destination node (B), a maximum hop time, a sliding window average of hop times over time and/or a success hop time, in which a predetermined percentage of hops is successfully performed.

11. The device according to any one of the preceding claims, wherein when receiving a data packet, the control unit (200) is adapted to adjust transmission parameters based on a distance the data packet has travelled.

12. The device according to one of the preceding claims, wherein the device (100) is used in telemanagement of a lighting system for switching on/off and/or controlling dimming patterns of luminaire nodes (10, 50), and/or reporting sensor data and/or luminaire status.

13. A system for controlling data packet transmissions in a wireless network, the system comprising:

a control centre (60); and

a plurality of nodes (10, 50), at least some of them comprising the device (100) according to one of the preceding claims; wherein data packets are transmitted from the sender node (A) to the destination node (B) via the wireless network.

14. The system according to claim 13, wherein the nodes (10, 50) are associated with luminaires of a lighting system.

15. A method for controlling data packet transmissions in a wireless network having a plurality of nodes (10, 50), the method comprising:

sending a data packet from a sender node (A) to a destination node (B); and adjusting an answer time-out of the data packet in the sender node (A) based on a distance between the sender node (A) and the destination node (B).