Merging GSM with WLAN in a mobile handset needn''t cause int

fly

Merging GSM with WLAN in a mobile handset needn''t cause int [Copy link]

When GSM is transmitting, the WLAN subsystem can''t receive WLAN packets (Fig. 1). Likewise, when GSM is in receive or monitor mode, the WLAN can'' transmit, as it will desensitize the GSM receiver. In addition, one radio chain must be turned off while the other transmits, due to interaction between the two transmitters. In most cases, the GSM transmitter will be given precedence and the WLAN transmitter will be disabled, as the existing infrastructure limits any changes to the GSM standard. What results is the need for some type of traffic management, or scheduling within the multi-mode solution. This is often achieved in the upper levels of the architecture. This scheduling, for instance, may exist within the application software or top-level baseband protocol stacks. While the result is a functional multi-mode solution, only one standard is ever instantaneously active at any time. As a consequence, only one of the two existing radio chains would ever be used at one time.

One vendor has developed multi-mode intellectual property (IP) that maximizes scheduling. By efficiently synchronizing GSM transmissions and receptions with those of WLAN, a single radio chain can be used for a multi-mode solution. This allows for a simple architecture, as the silicon content, complexity, component count, and overall size can be reduced. However, this IP has another advantage, as it reduces the overall time-averaged power consumption of the multi-mode handset.

To avoid desensing the GSM receiver, the IP schedules WLAN transmission at times when GSM doesn''t need the radio channel. Likewise, the scheduling algorithms can synchronize access point transmissions to GSM radio activity. This technology results in WLAN receptions that are never corrupted by GSM transmissions, or vice-versa.

While both handsets have two basebands and an application processor, the multi-mode handset needs only one and one antenna. This is due to the IP that allows one radio be shared by multiple standards concurrently (Fig. 2). The fundamental advantage the multi-mode IP is that it eliminates the interference between the WLAN and GSM. A second advantage is that basic scheduling and contention mechanisms are integrated into the radio. This why the GSM and WLAN baseband can share a radio transceiver with minimal changes to the baseband interfaces or upper level software and protocol stacks.

2. The interaction between the WLAN and GSM is shown using a proprietary architecture.

During GSM bursts, the WLAN system isn''t permitted to use the radio chain. Once the GSM finishes its radio activity, the WLAN subsystem can seize the radio, tune to the ISM band, and initiate a transmission or reception. Due to the turnaround times required to switch between the GSM and WLAN bands, the available time to transmit or receive packets for WLAN is less than that of a dual-radio solution. However, Quorum Systems'' has developed multi-mode synchronization IP that improves the handset''s performance with minimal impact to excess WLAN bandwidth. Using this architecture, three service links (GSM voice data, three-way VOIP conferencing, and two-way video conferencing) can be sustained simultaneously using one transceiver and two basebands.

Typically when implementing multi-mode IP, the probability of a successful WLAN transaction corresponds to the length of the WLAN packet. As the WLAN packets increase in duration, the likelihood that they will overlap with a competing GSM burst also increase. This will cause the WLAN packet to be dropped, requiring it to be retransmitted at a later time (Fig. 3). WLAN downlinks tend to be more robust as the WLAN receiver can operate during both GSM idle times and receive bursts.

3. This figure illustrates the performance degradation of a dual radio solution.

The Quorum Systems'' solution employs proprietary scheduling algorithms as the baseline reference. On the downlink, an access point transmits a data packet to the handset. Upon successful reception of the packet, the handset replies with an acknowledgement (ACK) burst transmission. This additional overhead is required when sustaining a link to any access point or base station. Because the ACK is of fixed length, the probability of it interfering with a GSM burst is independent of the data length. However, the probability of a successful data reception is inversely proportional to length of the data packet.

For short DATA packets, the probability of a successful data-ACK transaction is dominated by the ACK success. For longer transmissions, it''s dominated by the probability of a successful data packet reception. In the case of the uplink, a successful data-ACK transaction is almost solely dependent on the transmission''s length.

Because transmissions have a smaller contention window to deal with, the probability of a successful transmission, data-ACK falls off more quickly than the probability of a successful reception. These issues are somewhat mitigated be the fact that VoIP data traffic consists inherently of shorter data packets (on the order of 100 bytes or so). But even in this case, dual-radio solutions are still less optimal than those which employ a scheduling algorithm.

About the author
Lon Christensen is the CTO and co-founder of Quorum Systems, where he leads all technical and engineering functions within the company. Christensen can be reached at lon.christensen@quorumsystems.com.