Wireless Body Area Networks - Optical Zeitgeist Laboratory

Wireless Body Area Networks

IEEE 802.15.6 standard

Wireless Body Area Networks (WBANs) have emerged as a solution better suited for biological signal monitoring. They allow for mobility, usability, and comfort for the users. Furthermore, patients do not need to stay in hospitals to be monitored, which reduces health costs. These benefits have motivated the growth of several WBAN applications in medical and non-medical fields. For example, in gaming applications, the system can detect when the user is stressed and change the game level; in healthcare applications, the system can increase its data rate when an emergency situation (e.g., arrhythmia) occurs; in performance monitoring, the system can detect the worker’s stress level, which can be used for employers to improve the workspace conditions. Several research studies have been carried out to satisfy various application requirements and overcome challenges to guarantee good system performance. The main design goals of these studies have been to ensure reliability, low power consumption, and low packet loss. Toward this end, communications protocols used in WBANs such as Bluetooth and IEEE 802.15.4 WPAN have been investigated. As a result, it has been found that they are not enough flexible to satisfy all requirements of WBAN applications.

The aforementioned reasons have motivated the creation of a new WBAN standard known as IEEE 802.15.6 in November 2007. This standard is more flexible and can be used for both medical and non-medical applications, e.g., healthcare and entertainment. It covers the physical (PHY) and Medium Access Control (MAC) layers. The first one offers operation modes such as narrowband (NB), ultra-wideband (UWB), and human body communication (HBC). The second one offers different operation modes and medium access methods, which are explained below.

Network topology

A WBAN consists of a central node (coordinator or hub), which controls the WBAN operation and a number of nodes, which communicate directly with the hub. According to the standard, there are two network topologies for WBANs such as one-hop star and two-hop extended star. In the first one, the nodes exchange frames directly with the hub. In second one, the nodes and hub exchange frames through a relay-capable node, i.e., there is not a direct communication between them. These two possible network topologies are depicted in Fig. 1.

Figure 1. WBAN network topologies [1].

Reference model

As shown in Fig. 2, all nodes and hubs incorporate both PHY and MAC layers. The logical hub management entity (HME) and node management entity (NME) exchange management information between layers. For transmission, MAC service data units (MSDUs) are passed from the MAC client (higher layer) to the MAC sublayer through the MAC Service Access Point (SAP). Then, MAC protocol data units (MPDUs or MAC frames) are passed from the MAC layer to the PHY layer through the PHY SAP. Upon reception, MAC frames pass from the PHY layer to the MAC sublayer through the PHY SAP, and MSDUs from the MAC sublayer to the MAC client through the MAC SAP.

Figure 2. Reference model [1].

MAC layer

The standard defines three communication modes at the MAC layer for the hub, namely, (1) beacon mode with beacon periods (superframes), (2) non-beacon mode with superframes, and (3) non-beacon mode without superframes. For the communication modes using superframes, the standard divides the channel into superframes or beacon periods of equal length. These superframes contain allocation slots of equal duration used for data transmission.

Beacon mode with beacon periods (or superframes). The hub transmits beacon frames in active superframes to communicate the superframe boundaries. If there is no pending transmission, inactive superframes may exist after the active superframes. As illustrated in Fig. 3, the superframe structure is formed by a beacon frame (B), exclusive access phases (EAP1 and EAP2), random access phases (RAP1 and RAP2), as well as a managed access phase (MAP), B2 frame, and contention access phase (CAP). EAP1 and EAP2 are used for highest-priority traffic (emergency). RAP1, RAP2, and CAP are used for regular traffic. The MAP period is used for improvised, scheduled, and unscheduled access. The B2 frame is transmitted in order to provide a non-zero length CAP. Any of these periods can be disabled when the hub sets them to zero.

Figure 3. Beacon mode with beacon period [1].

EAPs, RAPs, and CAP periods use a contention-based channel access method. If the PHY operation mode is NB, the contention-based method used is carrier sense multiple access with collision avoidance (CSMA/CA). Whereas if the PHY operation mode is UWB, the contention-based channel access method is slotted ALOHA.
Non-beacon mode with superframes. The hub can access only the managed access phase (MAP) in any superframe without using beacons. Since a time reference is needed, the hub requires timed frames (T-poll) to inform about the superframe boundaries.

Figure 4. Non-beacon mode with superframes [1].
Non-beacon mode without superframes. In this mode, there are no beacon, superframes, allocated slots, and time reference. A node employs CSMA/CA or the unscheduled access method, whereby the hub sends poll or post frames at any time to the nodes. The poll frames are control frames used to grant polled allocation to the node in order to start frame transmissions. The post frames are management or data frames in order to let the hub start frame transmissions.

Figure 5. Non-beacon mode without superframes [1].

The above operation modes use different access methods, which are described next.

Random access
1. CSMA/CA access method. In order to get a new allocation, the node deploys a backoff counter and a contention window (CW). The backoff counter for each node is initialized to a random integer over the interval [0, CW]. The CW is selected from the interval (CWmin, CWmax) according to the user priority (UP) level (see Table 1). If the CW is small, there is a high probability to access the channel (emergency traffic). On the other hand, if the CW is large (e.g., 64), the probability to access the channel is low (regular traffic).
  
  Table 1. Contention window and contention probability bounds [1].
  
  The node considers a slot idle when it is idle from the start of the CSMA slot for the pCCATime period. The backoff counter is decremented by one every idle CSMA slot. When the backoff counter reaches zero, the node transmits the frame and the CW is configured as follows. It is set to CWmin, if the node did not get any allocation slot or if the frame transmission was successful. The CW is not changed, if the transmitter node does not require an ACK frame or if this is its m-th time where the node has failed consecutively, with m being odd. The CW is doubled if the node has failed consecutively n (even) times. If after doubling CW, it exceeds CWmax, the CW is set to CWmin.
  
  The backoff counter is locked by the node until the end of the current frame transmission if the channel is busy. It is also locked by the node, if the current time is outside of RAP and CAP for regular traffic or if the current time is outside of EAP, RAP, and CAP for emergency traffic. Moreover, when there is not enough time to finish the current transmission, the backoff counter is also blocked. On the other hand, the backoff counter is unlocked when the channel is idle for the pSIFS period within a CAP or RAP for regular traffic and when there is enough time to finish the current transmission.
  
  An example of contended allocations based on CSMA/CA is highlighted in Fig. 6. During the RAP1 period, the node waits for a SIFS period before it unlocks its backoff counter, which is selected randomly to be 3 with CW=8. The backoff counter is decremented by one in each idle slot. When the backoff counter reaches zero, the frame (F1) is transmitted. However, this transmission is assumed to fail (e.g., the ACK frame is not received), so the CW is not changed (odd number of failures) and the backoff counter is set to 5 and locked. During the CAP, the node waits for a SIFS period before it unlocks the backoff counter. When the backoff counter is equal to 2, it is locked because there is not enough time to finish the transmission between the end of the slot and the end of the CAP. During the RAP2 period, the backoff counter is unlocked and a contention failure occurs again. Thus, the CW is doubled (CW=16) because it is an even number of failures and the backoff counter is set to 8 and locked. The node waits again for another SIFS period and the backoff is unlocked. This time, there is a successful transmission, so the CW is reset to CWmin and the backoff counter is set to 2 and locked.
  
  Figure 6. CSMA/CA illustration [1].
2. Slotted ALOHA access method. The nodes access the channel with different traffic priorities (contention probabilities) according to Table 1. The contention probability (CP) is randomly selected from the interval [0, 1] and must be configured as follows. The CP is equal to CPmax, if the node did not get any contended allocation or a successful transmission was done. The CP is not changed, if the transmission did not require an ACK or if this is the m-th consecutively failed transmission, with m being odd. The CP is halved, if this is the n-th consecutively failed transmission, with n being even. If the CP value is smaller than CPmin (after it is halved), the CP value is set to CPmin. The slotted ALOHA access is depicted in Fig. 7. In RAP1, the node did not get a contended allocation, so the CP is set to CPmax. Then, a contended allocation is obtained and one frame is sent but it fails, so CP is not changed. In CAP, a transmission failure occurs again and the CP is halved (even number of transmission failures). In RAP2, the node finally can transmit and the CP is set to CPmax.
  
  Figure 7. Slotted ALOHA illustration [1].
Improvised access method. This access method is employed to send poll or post frames outside the scheduled allocations in beacon and non-beacon modes. The hub uses polls (control frames) to grant an immediate polled allocation to the node in order to start one or more transmissions. The posts (management frames) are used by the hub to send network information.
Scheduled access method. This access method allows the nodes to get scheduled time intervals for frame transmissions. A 1-periodic allocation let the node get one or more time intervals in every superframe. An m-periodic allocation let the node get one or more time intervals in every m-th superframe, which allows to put the node asleep between m superframes.

Improvements in IEEE 802.15.6 standard

The IEEE 802.15.6 standard helps increase energy efficiency and reliability. Toward this end, the standard uses the following mechanisms: i) sleep mode can be used for a long time, ii) beacon can be shifted with a known offset every superframe to minimize interference, iii) narrowband interference is avoided with dynamic channel hopping, iv) a single relay can be used when there is a bad link between one node and hub, so a direct communication between them is replaced with a communication through the relay node, v) there are 8 priority levels for different traffic types, vi) the CW is increased after each unsuccessful transmission with the backoff period selected between [0, CW], which helps reduce failed transmissions, vii) there are specific periods for high traffic (EAPs) and for regular traffic (RAPs and CAP), viii) the backoff count can be locked if there is not enough time to finish a transmission frame, ix) the BAN provides priority service, which is configured according to Table 2, and finally, x) three security levels are defined to ensure different levels of security such as level 0 (unsecured communication), level 1 (authentication but no encryption), and level 2 (authentication and encryption).

Table 2. BAN priority field encoding [1].

A more detailed discussion of the MAC techniques used in the IEEE 802.15.6 standard to improve reliability and efficiency is presented in [2]. The major findings of his discussion can be summarized as follows:

Dynamic slot allocation: Packet loss occurs due to outage periods, which consist of continuous periods of attenuation below the receiver sensitivity. Energy and time slots are wasted because of these periods. Therefore, the target is to avoid the outage state by increasing reliability without increasing energy consumption. The authors emphasized that if the link state is known, an effective slot allocation can be made to increase the reliability of the network. Thus, the slot allocation for the next superframe can be scheduled according to information about the link.
Scheduling of retransmissions: The authors proposed an adaptive retransmission technique that can be accomplished when the retransmission slots are allocated according to given channel conditions. Thus, retransmissions can be postponed by allocating slots at the end of the superframe in order to avoid outage conditions. Due to the fact that the extra energy managed in retransmissions has a limit, an efficient slot scheduling algorithm is needed.
Using relays: If the outage period is shorter than the packet delivery deadline retransmissions are effective. On the other hand, if the outage period is longer, a more effective solution is to use a relay around the bad link, at the expense of increased power consumption because of the relay.
Controlling transmit power: This option allows for improving reliability and energy efficiency. The channel attenuation is estimated by the transmitting device according to the previous packet transmission such that the transmit power can be adjusted depending on this information.

References