EP1845584A1

EP1845584A1 - Apparatus for selecting a beamforming direction

Info

Publication number: EP1845584A1
Application number: EP06112581A
Authority: EP
Inventors: Joerg Widmer; Imad Aad; Robert Vilzmann; Christian Hartmann
Original assignee: NTT Docomo Inc
Current assignee: NTT Docomo Inc
Priority date: 2006-04-12
Filing date: 2006-04-12
Publication date: 2007-10-17
Anticipated expiration: 2026-04-12
Also published as: DE602006004175D1; EP1845584B1; JP2007288786A; JP4629065B2

Abstract

Apparatus for selecting the beamforming direction of a node in a wireless network, said apparatus comprising:
a control module for controlling the beamforming direction of an antenna connected to said apparatus, said control module changing the beam angle in individual steps;
a receiving module for overhearing the ongoing transmissions and for extracting one or more neighborhood parameters being indicative of the overheard transmission at the different angle steps;
said control module calculating a decision parameter for each angle step and comparing said decision parameters of said different angle steps to select based thereupon the angle corresponding to the optimum decision parameter as the beamforming direction for said antenna.

Description

FIELD OF INVENTION

The present invention relates to the beamforming of antenna arrays.

BACKGROUND OF THE INVENTION

Using beam-antennas has proven to be a promising technique for wireless devices. A mobile device with a beam antenna may select the direction of its beam-pattern in a wireless network and may thereby positively affect connectivity and stability of the network. By beam direction we refer to the orientation of the antenna pattern, where the pattern can have arbitrary shape as determined by the antenna array. Beamforming may be applied, among others, for wireless access networks, mesh networks, vehicular networks and sensor networks.
Generally speaking there are two known approaches for beamforming, namely

a) Random Direction Beamforming (RDB), and
b) Communication-based Beamforming

While the first solution (approach a)) is simple with reduced overhead, its performance remains limited. Random direction beamforming (RDB) (which is e.g. described in C. Bettstetter, C. Hartmann and C. Moser, "How does randomized beamforming improve the connectivity of ad hoc networks?", In Proc. IEEE Intern. Conf. on Communications (ICC), Seoul, Korea, May 16-20, 2005) is a simple method that consists of pointing (and fixing) the beam of a directional-antenna in a random direction. RDB showed considerable improvement compared to using omni-directional antennas. However, this approach shows some limitations when nodes close to the border of the network or next to an obstacle beamform in a direction where they have no neighbors, as can be seen from Fig. 1 which illustrates the case where nodes x and y point their beams in a direction where no neighbors are located. Such nodes are isolated from the rest of the network.
The second solution (approach b) of using Communication-based Beamforming considerably increases the system performance, at the cost of high complexity. This approach is e.g. described in R. Roy Choudhury, X. Yang, R. Ramanathan and N. H. Vaidya, "Using directional antennas for medium access control in ad hoc networks", in ACM MobiCom 2002. It involves choosing the direction of the beam based on routing information. Communication-based beamforming forces nodes to continuously search for and adapt to the position of their communication partner, which requires tight coordination between nodes, resulting in high complexity and overhead. This means that the beamforming mechanism and the routing mechanism interact and are not independent of each other. The beamforming mechanism must be aware of routing information such as the identity of the communication partner (or at least the next hop) of the node and its location, which increases the complexity of the overall system. The individual nodes must be aware of the angle at which they can reach their communication partner, which can be obtained through angle-of-arrival estimation or using information about their position and the position of their immediate communication partners (next hop). This means that the (OSI) layer which is responsible for the routing and the layer which is responsible for the beamforming (e.g. the MAC layer) must interact, must exchange information and become interdependent. This situation is unfavorable in terms of complexity of the overall beamforming mechanism; moreover, its implementation in existing network infrastructures becomes difficult because it requires modifications at two different layers of the ISO-OSlmodel.
It is therefore an object of the invention to provide a considerable improvement over RDB which can be achieved without the complexity of "Communication-based beamforming".

SUMMMARY OF THE INVENTION

According to one embodiment there is provided an apparatus for selecting the beamforming direction of a node in an ad-hoc network, said apparatus comprising:

a control module for controlling the beamforming direction of an antenna connected to said apparatus, said control module changing the beam angle in individual steps;
a receiving module for overhearing the ongoing transmissions and for extracting one or more parameters being indicative of the overheard transmissions at the different angle steps;
said control module calculating a decision parameter for each angle step and comparing said decision parameters of said different angle steps to select the angle corresponding to the optimum decision parameter as the beamforming direction for said antenna based on said comparison.

According to this embodiment there may be implemented a beamforming antenna in wireless networks in order to enhance connectivity and robustness of routing and to decrease mutual interference.
According to one embodiment said neighborhood parameters are transmitted by other nodes without specific request from said node which is to select its beamforming angle, and wherein the weights with which said neighborhood parameters obtained from multiple nodes at different angles are considered for the calculation of the decision parameter are independent of the present or intended communication partner of the node which is to select its beamforming direction.
This means that the selection mechanism can be based on a "passive" overhearing of ongoing communications rather than on active requests which would involve additional communication. Moreover, rather than basing the selection on the present or intended communication partner of the node (i.e. its next hop) the selection is based on the overall neighborhood as reflected by the parameters derived from the overheard transmissions, thereby avoiding a frequent change of the angle and furthermore cascading effects which may arise from such a change in case of communication based beamforming which would take into account any change in communication partners.
According to one embodiment said neighborhood parameters are included in data transmissions or dedicated beacon messages at a layer different from or below the layers at which the routing protocol operates.
This has the advantageous effect that the beamforming selection becomes independent of the routing protocol and thereby avoids the complications which are caused by the interdependence between beamforming and routing in case of the communication based beamforming approach.
Specifically, the approach according to this embodiment is much less complex than "communication-based beamforming" since it requires no coordination among nodes, while significantly outperforming random beamforming direction at a comparable complexity.
In this manner the method provides better network connectivity, shorter paths (therefore shorter end-to-end delays), lower interference, lower battery consumption of mobile devices and increased network capacity.
According to one embodiment said neighborhood parameters are transmitted by said nodes at regular intervals, and wherein said neighborhood parameters comprise one or more of the following:

the number of neighbors,
the battery level,
the congestion level,
the channel quality,
the relative or absolute position; and

Thereby the node may compute the optimal beam direction using information about its neighborhood obtained from said parameters such as number of neighbors, their energy level, channel quality, congestion level, position, etc.
According to one embodiment said control module is adapted to carry out the following:

sweeping its neighborhood by turning its beam in steps of a predetermined angle;
overhearing the neighbors' transmissions of data packets and also of beacons, in case beacon messages are used;
keeping each angle direction for a certain period of time before moving to the next angle;
constructing based on a set P of status parameters including all received relevant neighbor parameters p_k, for k=1,...,m a decision parameter F(P);
if a beam direction results in an increase of F(P) over the previous one which is larger than a threshold, to use the new beam direction;
repeating the foregoing steps at regular intervals or additionally upon a manual trigger.

In this manner a beamforming selection algorithm may be implemented. The regular repetition ensures that the algorithm adapts to changes of the network conditions. Additionally there may be provided the possibility to manually trigger the beamforming selection, e.g. f a user or an operator considers that the network may have become unstable.
According to one embodiment said control module is adapted to adapt the incrementing step size of the angle and/or the step duration at a certain angle during the sweeping to obtain a statistically significant sample of the status updates in the corresponding direction.
This improves the reliability and efficiency of the beamforming selection algorithm.
According to one embodiment said neighborhood information is transmitted by said nodes piggy-packed on the normal traffic or by using a dedicated beacon message for transmitting said neighborhood information.
According to one embodiment said apparatus comprises a module for changing a threshold parameter which is applied to determine whether a change in the decision parameter is sufficient to change the beamforming direction, whereas said module is adapted to increase said threshold in case it detects that the network is too unstable, and/or
said module is adapted to decrease said threshold in case it detects that the network is too stable.
This enables said apparatus to adapt to the overall stability condition of the network.
According to one embodiment said apparatus comprises:

receiving said neighborhood parameters from neighbors being at a distance of one hop; and
calculating said decision parameter based on the neighborhood parameters received from said one-hop neighbors.

This makes it comparatively easy to implement the mechanism into existing network structures without significant modifications of the individual nodes because each node anyway transmits to some extent parameters (such as e.g. its MAC address) which may be used as neighborhood parameters.
According to one embodiment said apparatus comprises:

receiving said neighborhood parameters from neighbors being at a distance of more than one hop; and
calculating said decision parameter based on the neighborhood parameters received from said neighbors including the neighbors at a distance of more than one hop.

According to one embodiment said apparatus comprises:

determining said decision parameter based on the maximum number of neighbors detected at a certain angle.

This is a relatively easy way to determine metrics for calculating a decision parameter which yields a reasonably good result.
According to one embodiment there is provided a network comprising a plurality of nodes, each node comprising an apparatus according to one of the preceding claims, whereas said network is one of the following:

a sensor network;
a vehicular network;
a wireless mesh network;
a radio-access network.

According to one embodiment said apparatus comprises a module for randomly choosing the timing of the first beamforming selection carried out by this node, and for choosing the timing of subsequent beamforming selection operations at regular intervals.
This allows to avoid simultaneous execution of many beamforming selection operations which might negatively influence the stability of the network.
According to one embodiment said regular intervals vary in accordance with some random parameter chosen by said timing selection module.
This ensures that the selection is carried out regularly to ensure a continuous adaptation to changing circumstances, while the introduction of a random element avoids the clustering of selection operations at certain moments or periods of time.
According to one embodiment the selection operations are carried out on an ordered sequence. This ensures that two nodes do not simultaneously choose their beamforming direction. However, this requires node coordination to some extent.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 schematically illustrates a beamforming mechanism according to the prior art.
Fig. 2 schematically illustrates a configuration according to an embodiment of the invention,
Fig. 3 schematically illustrates a configuration according to a further embodiment of the invention.
Fig. 4 illustrates a pseudocode of an algorithm according to an embodiment of the invention.
Fig. 5A schematically illustrates a beamforming situation resulting from an embodiment of the invention.
Fig. 5B schematically illustrates a beamforming situation resulting from a further embodiment of the invention.

DETAILED DESCRIPTION

In the following the present invention will be described by exemplary embodiments in connection with the accompanying drawings.
According to a first embodiment there is provided a beamforming method and apparatus that improves performance compared to RDB, while avoiding the complexity of communication-based beamforming. In one embodiment the selection of the beamforming direction is based on aggregate information collected from neighboring nodes.
In the embodiment, some information can be gathered at each node without requiring any cooperation from other nodes (e.g. by overhearing ongoing transmissions and estimating the number of neighbors in each beam direction). The information about the number of neighbors may be derived from the overheard transmissions by identifying the number of different MAC (Media Access Control) addresses found in the overheard transmissions. This information (the MAC address which is a unique identifier attached to most forms of networking equipment) typically is contained in the data stream originating from a certain node, and therefore by identifying the number of different MAC addresses one can identify the number of different nodes. A control mechanism may cause the beam of the antenna to carry out a sweep in incrementing angles, and at each angle the number of different nodes is determined. The angle where the maximum number of nodes has been found may then be chosen as the direction of the beam for the antenna.
This means that according to one embodiment which is schematically illustrated in Fig. 2 there is provided a receiving module 210 which is connected to antenna 200 and which overhears the data transmitted from other nodes in incrementing angles, furthermore a control module 220 which is capable of controlling the beam direction of the antenna through control signal 230 and sweeps through the different angles (in incrementing steps) and extracts at each angle neighborhood parameters (according to one embodiment MAC addresses) from the overheard data stream, and which then based on the neighborhood parameters calculates a decision parameter for each angle. Based on a comparison of the decision parameters for the different angles the control module selects the angle to be chosen as the beamforming direction for the antenna. According to one embodiment this is the angle in which the highest number of nodes are located, in other words the angle at which the most distinct MAC addresses could be found.
According to one embodiment the receiving module and the control module may be implemented by a digital signal processor which is suitably programmed to operate as described before.
Fig. 3 illustrates a block diagram of a node according to a further embodiment of the invention and its operation. During the beam direction optimization phase, the transceiver gathers the overheard data and the (periodically transmitted) neighborhood parameters p_k. Those parameters (p_k) are used by the "computation of beam direction" module that controls the antenna beam direction. They are gathered using the current beam pattern without estimating their angle of arrival. During normal operation, each node periodically transmits its own status parameters p_i, either in beacon messages or piggy-backed onto data packets.
The node illustrated in Fig. 3 also transmits its own internal parameters p_i which are then received by other nodes as "neighborhood parameters" p_k and which thereby influence the selection of the beamforming direction at the other nodes. According to one embodiment the neighborhood parameters are transmitted in the data link layer, or in one specific embodiment in the MAC layer which is a sub-layer of the data link layer. This is indicated in Fig.3 by the element labeled MAC. From the MAC layer the data is passed on to higher layers as indicated in Fig.3. However, the neighborhood parameters p_k are not passed on to higher layers because they are transmitted at the MAC layer and therefore extracted from the data stream already at this level. Similarly, the internal parameters p_i are fed into the outgoing data stream at the MAC layer level.
Transmitting the neighborhood parameters at the MAC layer has the advantage that the whole beamforming selection operation is carried out at a layer which lies below the layers which are e.g. concerned with the routing algorithm. This means that the beamforming angle selection can be performed independently of the routing algorithm and thereby avoids the complications which are introduced by the interdependence between routing and beamforming in case of communication-based beamforming.
According to one embodiment the beamforming direction selection mechanism operates independent of any information about the communication partner of the node which has to select its beamform. This means that the node does not take into account its present communication partner, its present next hop in an ongoing communication session, and so on. Rather it only performs a "passive" monitoring of its neighborhood by overhearing ongoing transmissions at different angles. This is independent of its own communication intentions or conditions in the sense that the routing information such as its next hop and the location or direction of the next hop is not influencing the decision of the beamforming selection. Instead, rather than considering its own communication conditions and requirements a node scans its neighborhood with respect to the ongoing traffic to determine therefrom parameters indicative of the neighborhood as a whole (i.e. cumulative neighborhood parameters) rather than considering parameters reflecting its own situation and intention with respect to its communication partners. If any information from the present or intended communication partner of the node is considered at all, they are only considered equally with information from other nodes at the same angle step, i.e. there is no preference regarding any node the communication of which is overheard. In this manner the node does not need the extensive communication with other nodes to obtain relevant parameters for its own situation concerning its own communication requirements (such as determining "where is my next hop, at which angle in forward and in backward direction of the present or intended communication path") because instead the node rather than being concerned with its own communication partners is concerned about the overall situation of its neighborhood. The node obtains this information without actively "requesting" it but rather by "passively" overhearing it. This can be achieved by "passively" scanning the ongoing transmissions and extracting the neighborhood parameters contained in these transmissions. The neighborhood parameters are transmitted by the nodes of the neighborhood without specific request, e.g. in regular intervals using a piggy-back mechanism or a beacon signal.
According to one embodiment nodes can also include status parameters, e.g. their own energy level, the number of neighbors they have, etc., in their beacons or piggy-back it onto data packets. This information is then received by other nodes as "neighborhood parameters", and in one embodiment it may be used for determining or calculating the decision parameter. Thereby an improvement of the beamforming direction selection may be achieved. The more knowledge a node has about the network (or the 1-, 2- or n-hop neighbors) the higher the improvement it can achieve by directing its beam-antenna. However, if information not only about a certain node itself (i.e. internal parameters p_i) but also information about neighboring nodes of this node should be transmitted, complexity and overhead costs increase. This is because the information must flow over several hops in the network if a node feeds not only its own status information p_i but also information about its neighboring nodes into the outgoing data traffic or into a beacon message.
According to one embodiment status parameters p_i may contain one or more of the following:

Number of neighbors
Battery level
Channel quality (Signal to Noise Ratio)
Congestion level
Position (relative or absolute position)
etc.

According to one embodiment this information is sent at regular intervals, e.g. periodically, with period T_b, without affecting the beam direction of the node transmitting them (i.e., beacons are transmitted using the current beam configuration).
The number of neighbors may be considered as already described before by selecting the angle where the maximum number of neighbors are located.
The battery level may be considered by choosing an angle where there is a high battery lifetime for the nodes, in other words, an angle having the maximum cumulative energy level. This has the effect that it increases the network stability because an angle where due to the low energy one node after another will fail would have the effect that many communication paths cease to exist and new routing paths would have to be chosen.
According to one embodiment the nodes may be weighted based on their congestion level and based thereupon a cumulative congestion parameter may be determined for the different angles. Then an angle having a low congestion level may be selected.
Considering the position may be done such that it is tried to cover an area as completely as possible to leave no "white spots" where it would be difficult for a newly entering node to find a connecting node. For that purpose it would be necessary for the nodes to determine and transmit their absolute or relative positions so that this information may be extracted from the overheard traffic by a node which is to determine its beamforming angle.
According to one embodiment more than one neighborhood parameter may be taken into account when determining the decision parameters for the different angles. This may e.g. be done by determining an individual decision parameter based on each of the neighborhood parameters to be considered and to then calculate cumulated decision parameter based on the individual decision parameters. In such a case different weights may be applied to the different individual decision parameters, e.g. if energy is of particular concern the individual decision parameter based on the battery level may be assigned higher weight than e.g. the number of neighboring nodes.
According to one embodiment the multiple different neighbourhod parameters are not used to determine individually different decision parameters which are then combined, but rather a final decision parameter is directly obtained from different neighborhood parameters.
In the following there will be described in somewhat more detail an algorithm for selecting the beamforming angle according to an embodiment of the invention. The algorithm is performed by each node of the network.
Each node runs an algorithm to (re-)compute the beam direction (Algorithm 1) independently from the other nodes. Figure 4 illustrates an example of pseudocode for implementing this algorithm according to an embodiment of the invention. It starts sweeping its neighborhood by turning its beam in steps of angle_step. In each step it overhears the neighbors' transmissions of data packets and also of beacons, in case beacon messages are used. Each angle direction is kept for a time period of time_step, before the node moves to the next step. It then constructs a set P of status parameters including all received neighbor parameters p_k, for k=1,...,m. The sweeping loop searches for the beam direction that maximizes a decision parameter F(P), where F() combines all the relevant parameters p_k (battery level, SNR, number of neighbors, etc.). If the new beam direction results in a considerable increase (larger than threshold) of F() over the previous one, the new beam direction is kept. This search algorithm is repeated every search_period_duration.
In this case, these changes have to be taken into account in the calculation of F(P) (e.g., by dividing the calculated value by the factor (time_step / avg_step) where avg_time step is the average of all the individual time steps).
The average number of status updates (be they piggybacked or through dedicated beacons) over time determines the speed with which the sweeping for the computation of the beam direction can occur. In case data packets are sent very frequently, it may be sufficient to piggyback status information only onto some of them. In contrast, if very little or no traffic occurs, inferring an accurate image might take too much time without using dedicated beacon messages. The beacon frequency can be adapted such that the total number of updates a node can overhear per time step remains relatively constant.
According to one embodiment there is used a threshold parameter to decide whether a change in the decision parameter or optimization value opt_value is considered significant enough to actually change the beam direction. The lower the threshold, the closer the beam direction tracks the optimum direction. On the other hand, frequent changes in beam direction have an impact on the stability of the network. Changing the beam direction results in changes to the parameter values of the neighbors and this can cause them to also change their beam direction in turn. In this case, a wireless routing algorithm that is running on top of the proposed beam forming protocol would have to frequently adjust to the changes in the neighborhood, computing new routes and tearing down invalid old ones. Therefore, it is preferable to adapt the threshold parameter to the responsiveness of the routing protocol, as well as the network requirements concerning energy consumption, availability of alternate paths, etc.
This intelligent adjustment can be done through additional signaling between the beam direction protocol and higher layers such as the routing or even application layer. A higher layer may change the threshold (as well as the other parameters of the algorithm) in case it detects that the network is too unstable. This may be detected by measuring e.g. the average duration of a connection, and if it is considered too short the network may be considered too unstable. Another possibility would be to measure the number of broken connections within a certain time period, if it is too high the network may be regarded as unstable.
On the other hand, it may also indicate that the network is considered so stable that a threshold decrease would be acceptable. A similar mechanism for measuring network stability may be applied in this case.
In the following a beamforming technique according to one embodiment will be described in somewhat more detail.
In this embodiment the mobile terminals initially beamform in a random direction. Then each node sweeps the main lobe by incrementing its beamforming angle by a predefined amount. Upon completing the sweep, the node beamforms in the direction where the node degree was maximum (i.e. where it found the highest number of neighbors). If this maximum node degree occurred in more than one direction, one of these directions is picked at random. This process is repeated periodically to account for possible changes in the network topology.
In this case, F(P) corresponds to the number of distinct MAC source addresses overheard during each step of the sweeping phase. This mechanism will in the following be referred to as maximum node degree beamforming (MNDB).
In the following a beamforming technique according to a further embodiment will be described in somewhat more detail.
In some non-homogeneous topologies, MNDB can result in sub-optimal connectivity, where nodes point their beams such that they form clusters with strong connectivity within each of them, but few connections between different clusters. This is because a node tends to direct its beam into an angle where it can "directly" reach a maximum number of neighbors, "directly" here means without intermediate hops. However this is not in all cases the best solution, as can be seen from Fig. 5A which depicts a situation where node X can reach the maximum number of hops directly by directing its beam towards cluster A. However, in terms of overall connectivity it would be advantageous if node X would direct its beam towards cluster B because thereby a connection between these two clusters could be established, thereby increasing overall connectivity.
To overcome this problem, according to one embodiment nodes may use the Two-hop Node Degree Beamforming algorithm (TNDB) to maximize the number of distinct 1- and 2-hop neighbors, as will become apparent from the following.
In this embodiment a node not only transmits its own MAC address as status information or neighborhood information, but additionally also the MAC addresses of its next hop neighbors. It will be understood that this may require some amendment of the protocol at the MAC layer because while the own MAC address usually is transmitted together with any communication data, the MAC address of the next hop is normally not included. However, this may be achieved by keeping the MAC addresses received with incoming data and then feeding them again into an outgoing data stream as neighborhood parameter. Any node receiving these neighborhood parameters may then use all of those MAC addresses, the one of the originating node and the MAC adresses of that node's neighbors, as input parameters for calculating the decision parameter F(P).
In such a case, the individual status updates p_k contain the list of overheard MAC source addresses of node k, and F(P) calculates from this the number of distinct MAC addresses of the two-hop neighborhood, by summing up the p_k and removing duplicate MAC addresses. Applying this mechanism to the situation shown in Fig. 5B it becomes apparent that in such a case cluster B becomes advantageous over cluster A because through the two-hop connections it contains a higher number of reachable nodes, and accordingly node X will direct its beam to cluster B.
Using TNDB as illustrated in Fig. 5B, node X is connected to its neighbors in cluster A through its side-lobes, and it uses its main-lobe to reach further neighbors in cluster B, thus improving connectivity between the two clusters.
Compared to using omni-directional antennas and random direction beamforming, both MNDB and TNDB provide better network connectivity, shorter paths (therefore reducing end-to-end delays), and lower interference. These algorithms can be further improved by using additional information from neighbor nodes such as available energy, congestion level and channel quality. This reduces the cost of operating the networks while increasing user satisfaction (e.g. lower battery consumption of mobile devices, increased network capacity, etc.). Moreover, an even more distant neighborhood than the two-hop neighborhood (three hops, four hops, n hops) may be considered.
According to one embodiment the status information or neighborhood information are sent by each node periodically in a predetermined interval. The moment of the initial transmission may be chosen randomly, e.g. within a certain timing window after switching on the node. The subsequent transmissions (beacon or piggy backed) may be sent in a predetermined interval, according to one embodiment also this interval varies to some extent randomly.
Introducing the random character of the beamforming selection timing at the start and possibly also for subsequent selection operations reduces the possibility that beamforming selections are carried out simultaneously which might negatively affect network stability. On the other hand, the provision of a regular interval (which may vary to some extent randomly) within which a beamforming selection is made ensures that the network continuously and regularly adapts to changing situations which may arise through the movement of the nodes.
According to a further embodiment the sequence within which the individual nodes transmit is predetermined or fixed, in other words the order in which the individual nodes select their beamforming direction does not change. However, such a fixed sequence requires at least some coordination between the individual nodes, at least to establish the order. It is therefore somewhat more complex than having each node choosing its timing for its selection operation on its own while introducing some random element to ensure a more or less distributed execution over time of the selection process.
In the following some environments where embodiments of the invention may be applied will be explained.
Consider a sensor network where the network elements (sensors) use beam-antennas. Compared to sensors using omni-directional antennas, beam-antennas help increasing the transmission rates and ranges, saving energy, extending the sensors' lifetimes and better connecting the network. However, some sensors may be disconnected from the rest of the network due to the heterogeneity of the topology (topology borders, empty areas, etc.). Selecting the beamforming direction using embodiments of the invention helps all sensors to find the direction that better connects them to the (rest of the) network.
Another example where embodiments of the invention may be applied are vehicular networks.
Consider a network of vehicular nodes connected wirelessly using beam-antennas. Vehicles may exchange emergency information, traffic information etc., either using single-hop or multi-hop communication. Selecting the beamforming direction using embodiments of the invention helps a vehicle point its beam into the direction that increases its connectivity. It also helps adapting to the topology dynamics such that the vehicle does not point its beam to an "empty" area (e.g. a lake, side of the highway, etc.). Beamforming in general helps increasing the radio range, and selecting the beamforming direction using embodiments of the invention further helps choosing the "right" beam-direction to shorten end-to-end paths and reduce packet delays.
Another example where embodiments of the invention may be applied are Auto-configurable Radio-Access Networks.
In radio-access networks, an access point using a beam antennas increases its transmission range (and its receive gain) in a given direction(s). Selecting the beamforming direction using embodiments of the invention helps further adapting the orientation of the antenna beam(s) to cover more users, adapt to the dynamics of the topology (e.g. which conference room in a hotel is used) and increase user satisfaction.
Another example where embodiments of the invention may be applied are Auto-configurable Mesh Networks.
Extending the previous radio-access networks example, mesh network elements can use embodiments of the invention to improve network connectivity, reduce interference and adapt to occasional dynamics of the network.
It will be understood by a skilled person that the embodiments described above may be implemented by hardware, by software, or by a combination of software and hardware. The modules described in connection with embodiments of the invention may be as a whole or in part implemented by microprocessors or computers which are suitably programmed such as to act in accordance with the methods explained in connection with embodiments of the invention.

Claims

Apparatus for selecting the beamforming direction of a node in a wireless network, said apparatus comprising:
a control module for controlling the beamforming direction of an antenna connected to said apparatus, said control module changing the beam angle in individual steps;

a receiving module for overhearing the ongoing transmissions and for extracting one or more neighborhood parameters being indicative of the overheard transmission at the different angle steps;

said control module calculating a decision parameter for each angle step and comparing said decision parameters of said different angle steps to select the angle corresponding to the optimum decision parameter as the beamforming direction for said antenna based on said comparison.
The method of claim 1, wherein
said neighborhood parameters are transmitted by other nodes without specific request from said node which is to select its beamforming angle, and
wherein the weights with which said neighborhood parameters obtained from multiple nodes at different angles are considered for the calculation of the decision parameter are independent of the present or intended communication partner of the node which is to select its beamforming direction.
The apparatus of claim 1 or 2, wherein
said neighborhood parameters are included in data transmissions or dedicated beacon messages at a layer below the layers at which the routing protocol operates.
The apparatus of one of the preceding claims, wherein
said neighborhood parameters transmitted by said nodes comprise one or more of the following:
the number of neighbors,

the battery level,

the congestion level,

the channel quality,

the relative or absolute position, and
wherein if more than one neighborhood parameter is to be considered there is calculated a cumulative decision parameter for each angle based on the multiple neighborhood parameters.
The apparatus of one of the preceding claims, wherein said control module is adapted to carry out the following:
sweeping its neighborhood by turning its beam in steps of a predetermined angle;

overhearing the neighbors' transmissions of data packets and also of beacons, in case beacon messages are used;

keeping each angle direction for a certain period of time before moving to the next angle;

constructing based on a set P of status parameters including all received relevant neighbor parameters p_k, for k=1,...,m, a decision parameter F(P);

if a beam direction results in a increase of F(P) over the previous one which is larger than a threshold, to use the new beam direction;

repeating the foregoing steps in regular intervals or additionally upon a manual trigger.
The apparatus of one of the preceding claims, wherein said control module is adapted to adapt the step size of increment of the angle and/or the step duration at a certain angle during the sweeping to obtain a statistically significant sample of the status updates in the corresponding direction.
The apparatus of one of the preceding claims, wherein said neighborhood information is transmitted by said nodes piggy-packed on the normal traffic or by using a dedicated beacon message for transmitting said neighborhood information.
The apparatus of one of the preceding claims, comprising:
a module for changing a threshold which is applied to determine whether a change in the decision parameter is sufficient to change the beamforming direction, whereby said module is adapted to increase said threshold in case it detects that the network is too unstable, and/or

said module is adapted to decrease said threshold in case it detects that the network is too stable.
The apparatus of one of the preceding claims, comprising:
receiving said neighborhood parameters from neighbors being at a distance of one hop; and

calculating said decision parameter based on the neighborhood parameters received from said one-hop neighbors.
The apparatus of one of the preceding claims, comprising:
receiving said neighborhood parameters from neighbors being at a distance of more than one hop; and

calculating said decision parameter based on the neighborhood parameters received from said neighbors including the neighbors at a distance of more than one hop.
The apparatus of one of the preceding claims, comprising:
determining said decision parameter based on the maximum number of neighbors detected at a certain angle.
The apparatus of one of the preceding claims, comprising:
a module for randomly choosing the timing of the first beamforming selection carried out by this node, and

for choosing the timing of subsequent beamforming selection operations at regular intervals.
The apparatus of claim 12, wherein said regular intervals vary in accordance with some random parameter choosen by said timing selection module.
The apparatus of one of the preceding claims, comprising:
a timing selection module for establishing a sequence order within which the nodes of the network carry out their beamforming selection operation.
A network comprising a plurality of nodes, each node comprising an apparatus according to one of the preceding claims, where said network is one of the following:
a sensor network;

a vehicular network;

a wireless mesh network;

a radio-access network.